|
Citation |
- Permanent Link:
- https://ufdc.ufl.edu/UFE0042424/00001
Material Information
- Title:
- A Linear Input-Varying Framework for Modeling and Control of Morphing Aircraft
- Creator:
- GRANT,DANIEL THURMOND
- Place of Publication:
- [Gainesville, Fla.]
- Publisher:
- University of Florida
- Publication Date:
- 2011
- Language:
- english
- Physical Description:
- 1 online resource (200 p.)
Thesis/Dissertation Information
- Degree:
- Doctorate ( Ph.D.)
- Degree Grantor:
- University of Florida
- Degree Disciplines:
- Aerospace Engineering
Mechanical and Aerospace Engineering
- Committee Chair:
- Lind, Richard C
- Committee Members:
- Barooah, Prabir
Ifju, Peter Crisalle, Oscar D
- Graduation Date:
- 4/30/2011
Subjects
- Subjects / Keywords:
- Aircraft ( jstor )
Aircraft maneuvers ( jstor ) Aircraft wings ( jstor ) Damping ( jstor ) Eigenvectors ( jstor ) Inertia ( jstor ) Polynomials ( jstor ) Sine function ( jstor ) Trajectories ( jstor ) Velocity ( jstor ) AIRCRAFT -- CONTROL -- DYNAMICS -- MORPHING -- UNMANNED Mechanical and Aerospace Engineering -- Dissertations, Academic -- UF
- Genre:
- Electronic Thesis or Dissertation
born-digital ( sobekcm ) Aerospace Engineering thesis, Ph.D.
Notes
- Abstract:
- Morphing, which changes the shape and configuration of an aircraft, is being adopted to expand mission capabilities of aircraft. The introduction of biological-inspired morphing is particularly attractive in that highly-agile birds present examples of desired shapes and configurations. A previous study adopted such morphing by designing a multiple-joint wing that represented the shoulder and elbow joints of a bird. The resulting variable-gull aircraft could rotate the wing section vertically at these joints to alter the flight dynamics. This paper extends that multiple-joint concept to allow a variable-sweep wing with independent inboard and outboard sections. The aircraft is designed and analyzed to demonstrate the range of flight dynamics which result from the morphing. In particular, the vehicle is shown to have enhanced crosswind rejection which is a certainly critical metric for the urban environments in which these aircraft are anticipated to operate.
Mission capability can be enabled by morphing an aircraft to optimize its aerodynamics and associated flight dynamics for each maneuver. Such optimization often consider the steady-state behavior of the configuration; however, the transient behavior must also be analyzed. In particular, the time-varying inertias have an effect on the flight dynamics that can adversely affect mission performance if not properly compensated. These inertia
terms cause coupling between the longitudinal and lateral-directional dynamics even for maneuvers around trim. A simulation of a variable-sweep aircraft undergoing a symmetric morphing for an altitude change shows a noticeable lateral translation in the flight path because of the induced asymmetry.
The flight dynamics of morphing aircraft must be analyzed to ensure shape-changing trajectories have the desired characteristics. The tools for describing flight dynamics of fixed-geometry aircraft are not valid for time-varying systems such as morphing aircraft. This paper introduces a method to relate the flight dynamics of morphing aircraft by interpreting a time-varying eigenvector in terms of flight modes. The time-varying eigenvector is actually defined through a decomposition of the state-transition matrix and thus describes an entire response through a morphing trajectory. A variable-sweep aircraft is analyzed to demonstrate the information that is obtained through this method and how the flight dynamics are altered by the time-varying morphing.
Also, morphing vehicles have inherently time-varying dynamics due to the alteration of their configurations; consequently, the numerous techniques for analysis and control of time-invariant systems are inappropriate. Therefore, a control scheme is introduced that directly considers a concept of time-varying pole to command morphing. The resulting trajectory minimizing tracking error for either a state response or a pole response. ( en )
- General Note:
- In the series University of Florida Digital Collections.
- General Note:
- Includes vita.
- Bibliography:
- Includes bibliographical references.
- Source of Description:
- Description based on online resource; title from PDF title page.
- Source of Description:
- This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
- Thesis:
- Thesis (Ph.D.)--University of Florida, 2011.
- Local:
- Adviser: Lind, Richard C.
- Statement of Responsibility:
- by DANIEL THURMOND GRANT.
Record Information
- Source Institution:
- University of Florida
- Holding Location:
- University of Florida
- Rights Management:
- Copyright GRANT,DANIEL THURMOND. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
- Resource Identifier:
- 755927399 ( OCLC )
- Classification:
- LD1780 2011 ( lcc )
|
Downloads |
This item has the following downloads:
|
Full Text |
PAGE 1
ALINEARINPUT-VARYINGFRAMEWORKFORMODELINGANDCONTROLOFMORPHINGAIRCRAFTByDANIELT.GRANTADISSERTATIONPRESENTEDTOTHEGRADUATESCHOOLOFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENTOFTHEREQUIREMENTSFORTHEDEGREEOFDOCTOROFPHILOSOPHYUNIVERSITYOFFLORIDA2011 1
PAGE 2
c2011DanielT.Grant 2
PAGE 3
\IcandoallthingsthroughChristwhichstrengthensme" Philippians4:13 Dedicatedwithlovetomywifeandfamily 3
PAGE 4
ACKNOWLEDGMENTS IwouldrstliketoacknowledgetheUniversityofFloridaandUnitedStatesAirForceforsupportingmyambitionandgivingmetheopportunitytoconductsuchresearch.ThanksshouldbegiventoDr.WarrenDixon,Dr.PrabirBarooah,Dr.PeterIfjuandDr.OscarCrisalleforprovidingdirectionandservingasmycommitteemembers.IwouldalsoliketothankmyseniorlabfellowsDr.MujahidAbdulrahim,Dr.AdamWatkins,Dr.RyanCausey,Dr.JosephKehoeandDr.SeanRegisfordfortheirpatienceandguidancewhilementoringme.MuchthanksisgiventomycolleaguesSankethBhat,BrianRoberts,RobertLove,BaronJohnson,RyanHurley,AbePachikara,DongTranandStevenSorelyfortheirsupport,inspiration,andperseverance.IwouldliketoextendmysincerestthanksandgratitudetoDr.RickLindforhiseortsinsupportingmyeducation,guidanceinacademia,andprovidingmewithaninvaluableopportunitytoachievesuccess.Withoutthecontinuoussupportandunconditionalloveofmyfamilyandfriends,noneofthisworkwouldhavebeenpossible.Lastandmostimportant,IwouldliketothankmylovingwifeforbeingtheshininglightbehindallthatIdo. 4
PAGE 5
TABLEOFCONTENTS page ACKNOWLEDGMENTS ................................. 4 LISTOFTABLES ..................................... 9 LISTOFFIGURES .................................... 11 ABSTRACT ........................................ 17 CHAPTER 1INTRODUCTION .................................. 19 1.1Motivation .................................... 19 1.2ProblemDescription .............................. 20 1.3ProblemStatement ............................... 23 1.3.1Contributions .............................. 24 1.3.2Papers .................................. 25 1.4DissertationOverview ............................. 28 2EQUATIONSOFMOTION ............................. 29 2.1AircraftAxisSystem .............................. 29 2.1.1BodyAxisSystem ............................ 29 2.1.2StabilityAxisSystem .......................... 29 2.1.3EarthAxisSystem ........................... 30 2.2CoordinateTransformations .......................... 31 2.2.1EarthtoBodyFrame .......................... 31 2.2.2StabilitytoBodyFrame ........................ 33 2.3NonlinearEquationsofMotion ........................ 34 2.3.1DynamicEquations ........................... 34 2.3.1.1Forceequations ........................ 34 2.3.1.2Momentequations ...................... 39 2.3.2KinematicEquations .......................... 43 2.3.2.1Orientationequations .................... 43 2.3.2.2Positionequations ...................... 44 2.3.3TheEquationsCollected ........................ 45 2.4LinearizedEquationsofMotion ........................ 46 2.5Examples .................................... 49 2.5.1Linearization ............................... 50 2.5.2AsymmetricMorphing ......................... 53 2.5.3SymmetricConguration ........................ 54 2.6ATechnicalApproachtotheEquationsofMotion .............. 55 5
PAGE 6
3LINEARTIME-VARYINGEIGENSTRUCTURESANDTHEIRSTABILITY:ASURVEY ...................................... 57 3.1DenitionofanLTVSystem .......................... 57 3.2Kamen'sConceptofPolesofanLTVSystem ................. 57 3.2.1TwoStateSystem ............................ 57 3.2.2FourStateSystem ............................ 58 3.3ZhuandJohnson ................................ 60 3.3.1GeneralizedPD-Eigenvectors ...................... 62 3.3.2StabilityCriteria ............................ 63 3.4Wu ........................................ 64 3.4.1Formulation ............................... 64 3.4.2StabilityCriteria ............................ 66 3.5O'BrienandIglesias .............................. 67 3.5.1Formulation ............................... 67 3.5.2StabilityCriteria ............................ 69 3.5.3SpecialCases .............................. 71 3.6MethodologyComparisons ........................... 71 4LINEARTIME-VARYINGMODALANALYSIS ................. 74 4.1SystemDynamics ................................ 74 4.1.1LinearTime-InvariantSystems:Formulation ............. 74 4.1.2LinearTime-InvariantSystems:Response ............... 75 4.1.3LinearTime-VaryingSystems:Formulation .............. 76 4.1.4LinearTime-VaryingSystems:Response ............... 77 4.2ModalInterpretation:Kamen ......................... 78 4.2.1Damping ................................. 79 4.2.2Frequency ................................ 80 4.3ModalInterpretation:O'Brien ......................... 81 4.3.1DampingandNaturalFrequency .................... 81 4.3.2ModeShapes .............................. 82 4.4Example:SimpleMechanicalSystem ..................... 83 5CONTROLDESIGN ................................. 94 5.1InherentClosed-LoopNonlinearity ...................... 94 5.2Quasi-StaticApproach ............................. 95 5.2.1Synthesis ................................. 95 5.2.2Example:Gull-WingedAircraft .................... 96 5.3StabilizingControl:DisturbanceRejection .................. 97 5.3.1Synthesis ................................. 97 5.3.2Example:Chord-VaryingMorphing .................. 100 5.4FeedForwardOptimalControl ......................... 102 5.4.1State-ResponseTracking ........................ 102 5.4.2Pole-ResponseTracking ......................... 103 6
PAGE 7
5.4.3Example:Mass-SpringSystem .................... 104 5.4.3.1System ............................. 104 5.4.3.2MorphingBasis ........................ 104 5.4.3.3TrackingaSystemResponse:PolynomialMorphing .... 106 5.4.3.4TrackingaSystemResponse:Piecewise-PolynomialMorphing 107 5.4.3.5TrackingaPoleResponse:PolynomialMorphing ..... 108 5.4.3.6TrackingaPoleResponse:Piecewise-PolynomialMorphing 110 5.5H1FeedbackControl:WithoutInertialEects ............... 112 5.6FutureWorkandChallenges .......................... 117 6EXAMPLEOFVARIABLESWEEPAIRCRAFT ................. 120 6.1Design ...................................... 120 6.1.1BiologicalInspiration .......................... 120 6.1.2MechanicalDesign ............................ 121 6.1.3TechnicalSpecications ......................... 123 6.2Modeling ..................................... 125 6.2.1ComputationalTools .......................... 125 6.2.2SweepDetermination .......................... 127 6.3AerodynamicProperties ............................ 128 6.3.1SymmetricCongurations:Aerodynamics ............... 128 6.3.2SymmetricCongurations:FlightDynamics ............. 130 6.3.3AsymmetricCongurations ....................... 133 6.3.3.1Flightdynamics ........................ 135 6.3.3.2Modecharacterization .................... 136 6.3.3.3Crosswindrejection ...................... 138 6.4DynamicProperties ............................... 140 6.4.1MissionScenario ............................. 140 6.4.1.1Divemaneuver ........................ 141 6.4.1.2Turnmaneuver ........................ 141 6.4.2MassDistribution ............................ 142 6.4.3ManeuverAssumptions ......................... 143 6.4.4DiveManuever ............................. 144 6.4.4.1Modeling ........................... 144 6.4.4.2Altitudecontroller ...................... 145 6.4.4.3Time-varyingdynamics .................... 146 6.4.4.4Simulation ........................... 147 6.4.4.5Missionevaluation ...................... 149 6.4.4.6Eectsoftime-varyinginertia ................ 150 6.4.5CoordinatedTurnManeuver ...................... 152 6.4.5.1Modeling ........................... 152 6.4.5.2Turncontroller ........................ 153 6.4.5.3Time-varyingdynamics .................... 154 6.4.5.4Simulation ........................... 155 6.4.5.5Missionevaluation ...................... 159 7
PAGE 8
6.4.5.6Eectsoftime-varyinginertia ................ 160 6.5Time-VaryingModalAnalysis ......................... 161 6.5.1Kamen'sMethod ............................ 161 6.5.1.1Statespacemodel ...................... 161 6.5.1.2Morphingtrajectory ..................... 163 6.5.1.3Time-varyingpoles ...................... 163 6.5.1.4Modalinterpretation ..................... 167 6.5.2O'Brien'sMethod ............................ 169 6.5.2.1Statespacemodel ...................... 169 6.5.2.2Time-varyingpolesandstabilitymodes ........... 172 6.6FeedforwardControl .............................. 178 6.6.1MorphingBasis ............................. 178 6.6.2TrackingASystemResponse ...................... 182 6.6.2.1Polynomialmorphing ..................... 182 6.6.2.2Piecewise-polynomialmorphing ............... 183 6.6.3TrackingaPoleResponse ........................ 184 6.6.3.1Polynomialmorphing ..................... 184 6.6.3.2Piecewise-polynomialmorphing ............... 186 7CONCLUSION .................................... 188 REFERENCES ....................................... 190 BIOGRAPHICALSKETCH ................................ 200 8
PAGE 9
LISTOFTABLES Table page 1-1Publicationsresultingfromresearch ......................... 27 4-1Mass-spring-dampersystemparameters ....................... 84 5-1Gainsforchord-varyingaircraftdisturbancerejectioncontroller ......... 100 5-2Boundsoncoecientofmorphingbasis ...................... 106 5-3Optimalpolynomialmorphingtotrackdesiredpolynomialmorphing ...... 106 5-4Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing ....... 106 5-5Optimalpiecewise-polynomialmorphingtotrackdesiredsinusoidalmorphing .. 108 5-6Optimalpolynomialmorphingtotrackdesiredpolynomialmorphing ...... 109 5-7Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing ....... 109 5-8Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing ....... 111 6-1Empennagespecications .............................. 124 6-2Referenceparametersforsymmetricsweep ..................... 125 6-3Setofeigenvalues ................................... 137 6-4Timeconstantsofnon-oscillatorymodes ...................... 137 6-5Modeshapesofnon-oscillatorymodes ....................... 138 6-6Modalpropertiesofoscillatorymodes ........................ 138 6-7Modeshapesofoscillatorymodes .......................... 139 6-8Individualpointmasses ............................... 143 6-9Characteristicsofelementsgivenascentroidposition(in)andmomentsofinertia(gin2) ........................................ 143 6-10Boundsoncoecientofmorphingbasis ...................... 181 6-11Optimalpolynomialmorphingtotrackdesiredpolynomialmorphing ...... 182 6-12Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing ....... 182 6-13Optimalpiecewise-polynomialmorphingtotrackdesiredsinusoidalmorphing .. 184 6-14Optimalpolynomialmorphingtotrackdesiredpolynomialmorphing ...... 185 6-15Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing ....... 185 9
PAGE 10
6-16Optimalpiecewise-polynomialmorphingtotrackdesiredsinusoidalmorphing .. 187 10
PAGE 11
LISTOFFIGURES Figure page 1-1Surveillancemissionthroughanurbanenvironment ................ 19 1-2Vision-basedpathplanning ............................. 20 1-3Readinessformissioncapability ........................... 21 1-4MorphingMAVs:A)CapableofhorizontalmorphingB)Capableofverticalmorphing ....................................... 21 2-1Body-FixedCoordinateFrame ............................ 29 2-2StabilityCoordinateFrame ............................. 30 2-3Earth-FixedCoordinateFrame ........................... 30 2-4Rotationthrough .................................. 31 2-5Rotationthrough .................................. 32 2-6Rotationthrough .................................. 33 2-7AsymmetricCongurations ............................. 53 2-8SymmetricCongurations .............................. 54 4-1Mass-Spring-DamperSystem ............................ 84 4-2Time-VaryingResponses:A)Non-Inertal:State1(|),State2()-356()-356()]TJ /F1 11.955 Tf 36.42 0 Td[()B)Inertial:State1(|),State2()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[() ....................... 86 4-3Time-VaryingModes:A)Non-Inertal:Mode1(|),Mode2(\000)]TJ /F1 11.955 Tf 27.9 0 Td[()B)Inertial:Mode1(|),Mode2()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[() ........................... 87 4-4Time-VaryingPoles:A)Non-Inertal:Pole1(\000)]TJ /F1 11.955 Tf 27.9 0 Td[(),Pole2(\001)]TJ /F1 11.955 Tf 21.92 0 Td[(),RealportionofdiscreteLTIpoles(|),Poleaverage()B)Inertial:Pole1()-267()-267()]TJ /F1 11.955 Tf 34.29 0 Td[(),Pole2()-222()]TJ /F1 11.955 Tf 27.23 0 Td[(),RealportionofdiscreteLTIpoles(|),Poleaverage() ....... 88 4-5Non-InertialTime-VaryingEigenvectors:A)Eigenvector1:(1,1)(|),(2,1)(\000)]TJ /F1 11.955 Tf 9.3 0 Td[()B)Eigenvector2:(1,1)(|),(2,1)()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() .................. 89 4-6InertialTime-VaryingEigenvectors:A)Eigenvector1:(1,1)(|),(2,1)()-119()-118()]TJ /F1 11.955 Tf 30.74 0 Td[()B)Eigenvector2:(1,1)(|),(2,1)()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() .................... 90 4-7Non-InertialQcolumnvectorvalues:A)Q11(|),Q21()-278()-278()]TJ /F1 11.955 Tf 34.55 0 Td[()B)Q12(|),Q22()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() ..................................... 90 4-8Qcolumnperiodicity:Q11(|),Q21()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[() ................... 91 11
PAGE 12
4-9Qcolumnfrequency:ImaginarypartofdiscreteLTIpole(|),1Qcolumnperiodicity(\000)]TJ /F1 11.955 Tf 9.3 0 Td[() ........................................... 91 4-10InertialQcolumnvectorvalues:A)Q11(|),Q21(\000)]TJ /F1 11.955 Tf 27.9 0 Td[()B)Q12(|),Q22(\000)]TJ /F1 11.955 Tf 9.3 0 Td[() ........................................... 92 4-11Qcolumnperiodicity:Q11(|),Q21()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[() ................... 93 4-12Qcolumnfrequency:ImaginarypartofdiscreteLTIpole(|),LTVapproximatedfrequency()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() .................................. 93 5-1Model-FollowingSystem ............................... 97 5-2SimulationsoftheGull-MorphingAircraftModel ................. 97 5-3Chord-VaryingMAVModel(m) ........................... 100 5-4AngleofAttackResponseandAssociatedChordofAircraft ........... 101 5-5ResponsestoChord-VaryingMorphing ....................... 102 5-6ArchitectureforState-ResponseTracking ...................... 103 5-7ArchitectureforPole-ResponseTracking ...................... 104 5-8Mass-Spring-DamperSystem ............................ 104 5-9DesiredResponse(|)andOptimalResponse(\000)]TJ /F1 11.955 Tf 27.89 0 Td[()forA)First-OrderMorphingandB)Second-OrderMorphingandC)Third-OrderMorphing .......... 107 5-10DesiredResponse(|)andOptimalResponse()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()forSinusoidalMorphing 108 5-11DesiredResponse(|)andOptimalResponse()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()forSinusoidalMorphing 109 5-12DesiredResponse(|)andOptimalResponse(\000)]TJ /F1 11.955 Tf 27.89 0 Td[()forA)First-OrderMorphingandB)Second-OrderMorphingandC)Third-OrderMorphing .......... 110 5-13DesiredResponse(|)andOptimalResponse()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()forSinusoidalMorphing 111 5-14DesiredResponse(|)andOptimalResponse()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()forSinusoidalMorphing 111 6-1PicturesofSeagulls .................................. 120 6-2JointsonWing .................................... 121 6-3FloatingElbowJoint ................................. 122 6-4Feather-LikeElements ................................ 122 6-5Trackandrunnersystem ............................... 123 6-6Underwingsparstructure .............................. 124 12
PAGE 13
6-7Modelingoftheliftvectors ............................. 125 6-8Modelingofthetrailinglegvectors ......................... 126 6-9SweepCongurations ................................. 127 6-10SweepAngles ..................................... 128 6-11VariationofLiftwithAngleofAttackforSymmetricSweep ........... 129 6-12VariationofPitchMomentwithAngleofAttackforSymmetricSweep ..... 129 6-13VariationofRollMomentwithRollRateforSymmetricSweep .......... 130 6-14VariationofYawMomentwithAngleofSideslipforSymmetricSweep ..... 130 6-15Modeltostate-spaceowchart ........................... 131 6-16NumberofUnstablePolesofLongitudinalDynamicsforSymmetricSweep ... 131 6-17NumberofUnstablePolesofLateral-DirectionalDynamicsforSymmetricSweep 132 6-18NumberofOscillatoryPolesforLongitudinalDynamicswithSymmetricSweep 132 6-19NumberofOscillatoryPolesforLateral-DirectionalDynamicswithSymmetricSweep ......................................... 133 6-20VariationofLiftwithAngleofAttackforAsymmetricSweep ........... 134 6-21VariationofPitchMomentwithAngleofAttackforAsymmetricSweep ..... 134 6-22VariationofRollMomentwithRollRateforAsymmetricSweep ......... 134 6-23VariationofYawMomentwithAngleofSideslipforAsymmetricSweep ..... 135 6-24VariationofCoupledAerodynamicsforAsymmetricSweep ............ 135 6-25NumberofUnstablePolesforDynamicswithAsymmetricSweep ........ 136 6-26NumberofOscillatoryPolesforDynamicswithAsymmetricSweep ....... 136 6-27EectiveAnglesofSideslip .............................. 139 6-28MaximumAngleofSideslipatwhichAircraftcanTrim .............. 140 6-29PointMassLocations ................................. 143 6-30ChangeinVelocityBasedonSymmetricMorphing:0deg( {2{ ),5deg( {/{ ),10deg( {{ ),15deg( {{ ),20deg( {.{ ),25deg( {{ ),30deg( {4{ ) ...... 144 6-31Closed-LoopBlockDiagram ............................. 146 6-32PlantModelwithTrimLogic ............................ 147 13
PAGE 14
6-33MorphingCongurationforFastMorphing( | ),SlowMorphing( )-222()-222()]TJ ET 0 0 1 RG 0 0 1 rg BT /F1 11.955 Tf 398.84 -11.96 Td[() .... 148 6-34AltitudeinResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( \000)]TJ ET 0 0 1 RG 0 0 1 rg BT /F1 11.955 Tf 36.22 -50.31 Td[(),FixedStraight( )-221(\001 ) .............................. 149 6-35PitchAngle(left)andPitchRate(right)inResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( )-221()-223()]TJ ET 0 0 1 RG 0 0 1 rg BT /F1 11.955 Tf 239.46 -88.66 Td[(),FixedStraight( )-222(\001 ) ........ 149 6-36AltitudeinResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( \000)]TJ ET 0 0 1 RG 0 0 1 rg BT /F1 11.955 Tf 36.22 -127.02 Td[(),FixedStraight( )-221(\001 ) .............................. 150 6-37AltitudeinResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( \000)]TJ ET 0 0 1 RG 0 0 1 rg BT /F1 11.955 Tf 36.22 -165.37 Td[(),FixedStraight( )-221(\001 ) .............................. 151 6-38AltitudeinResponsetoFastMorphing( | ),SlowMorphing( ::: ),FastMorphingWithoutInertia( )-222(\001 ),SlowMorphingWithoutInertia( )-222()-222()]TJ ET 0 0 1 RG 0 0 1 rg BT /F1 11.955 Tf 363.15 -203.72 Td[() ........ 152 6-39ChangeinVelocityBasedonAsymmetricMorphing:0deg( {2{ ),5deg( {/{ ),10deg( {{ ),15deg( {{ ),20deg( {.{ ),25deg( {{ ),30deg( {4{ ) ...... 153 6-40Open-LoopBlockDiagram .............................. 154 6-41PlantModelwithTrimLogic ............................ 155 6-42MorphingCongurationforFastMorphing( | ),SlowMorphing( )-222()-222()]TJ ET 0 0 1 RG 0 0 1 rg BT /F1 11.955 Tf 398.84 -313.8 Td[() .... 156 6-43TurninResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( \000)]TJ ET 0 0 1 RG 0 0 1 rg BT /F1 11.955 Tf 36.22 -352.15 Td[(),FixedStraight( )-169(\001 ):A)ThecompleteturnproleB)Theturnproleat270degreesC)Theturnproleat180degrees ................... 157 6-44RollAngle(left)andRollRate(right)inResponsetoFastMorphing(|),SlowMorphing()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() .................................. 157 6-45YawAngle(left)andYawRate(right)inResponsetoFastMorphing(|),SlowMorphing()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() .................................. 158 6-46PitchAngle(left)andPitchRate(right)inResponsetoFastMorphing(|),SlowMorphing()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() ............................... 158 6-47TurninResponsetoFastMorphing(|),SlowMorphing(:::),FixedSwept(\000)]TJ /F1 11.955 Tf 9.3 0 Td[(),FixedStraight()-221(\001) .............................. 159 6-48TurninResponsetoFastMorphing(|),SlowMorphing(:::),FastMorphingWithoutInertia()-222(\001),SlowMorphingWithoutInertia()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() ........ 160 6-49LongitudinalStatesduringMorphingfrom+30degto0degover1sec:ForwardVelocity(upperleft),VerticalVelocity(upperright),PitchRate(lowerleft),PitchAngle(lowerright) ............................... 164 14
PAGE 15
6-50LinearTime-VaryingPoles(|)andLinearTime-InvariantPoles()-20()-21()]TJ /F1 11.955 Tf 28.39 0 Td[()duringMorphingfrom+30degto0degover1sec:RealPart(upperleft)andImaginaryPart(lowerleft)ofp41andp42,RealPart(upperright)andImaginaryPart(lowerright)ofp43andp44 ................................. 165 6-51ModesAssociatedwithTime-VaryingPolesduringMorphingfrom+30degto0degover1sec:RealPart(upperleft)andImaginaryPart(lowerleft)of41and42,RealPart(upperright)andImaginaryPart(lowerright)of43and44 166 6-52NormalizedEigenvectorsAssociatedwithTime-VaryingModesduringMorphingfrom+30degto0degover1sec:Magnitude(upperleft)andPhase(lowerleft)ofv1andMagnitude(upperright)andPhase(lowerright)ofv2 ......... 167 6-53NaturalFrequencyAssociatedwithLinearTime-VaryingPoles(|)andLinearTime-InvariantPoles()-94()-95()]TJ /F1 11.955 Tf 30.16 0 Td[()duringMorphingfrom+30degto0degover1sec:Poles1and2(left)andPoles3and4(right) ................... 168 6-54EnvelopeAssociatedwithLinearTime-VaryingPoles(|)andLinearTime-InvariantPoles()-257()-257()]TJ /F1 11.955 Tf 34.04 0 Td[()duringMorphingfrom+30degto0degover1sec:Poles1and2(left)andPoles3and4(right) .......................... 169 6-55DampingRatioAssociatedwithLinearTime-VaryingPoles(|)andLinearTime-InvariantPoles()-257()-257()]TJ /F1 11.955 Tf 34.04 0 Td[()duringMorphingfrom+30degto0degover1sec:Poles1and2(left)andPoles3and4(right) .......................... 169 6-56Stabilitymodechangevstimechange:10degrees(|),20degrees()-258()-259()]TJ /F1 11.955 Tf 34.08 0 Td[(),30degrees(---)A)2secondsB)5secondsC)10secondsD)20seconds ..... 173 6-57Stabilitymodechangevsdegreechange:20seconds(|),10seconds()-216()-216()]TJ /F1 11.955 Tf 33.06 0 Td[(),5seconds(---),2seconds()A)10degreesB)20degreesC)30degrees .... 174 6-58Polechangevstimechange:10degrees(|),20degrees()-327()-328()]TJ /F1 11.955 Tf 35.73 0 Td[(),30degrees(---)A)2secondsB)5secondsC)10secondsD)20seconds .......... 175 6-59Polechangevsdegreechange:20seconds(|),10seconds()-281()-283()]TJ /F1 11.955 Tf 34.64 0 Td[(),5seconds(---),2seconds()A)10degreesB)20degreesC)30degrees ......... 176 6-60Longitudinalaircraft30degree,10secondmorphA)polesets:polesets1and2()-271()-271()]TJ /F1 11.955 Tf 34.38 0 Td[(),polesets3and4(---),time-invariantfrozentimeeigenvalues(|)B)stabilitymodes:mode1(|)mode2()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()mode3(---)mode4() .. 177 6-61LongitudinalaircraftQcolumnvectors1and2for30degree,10secondmorph:A)Q11(|)Q21()-252()-252()]TJ /F1 11.955 Tf 33.93 0 Td[()Q31(---)Q41()B)Q12(|)Q22()-252()-252()]TJ /F1 11.955 Tf 33.93 0 Td[()Q32(---)Q42() .................................... 178 6-62LongitudinalaircraftQcolumnvectors3and4for30degree,10secondmorph:A,B)Q13(|)Q23()-256()-256()]TJ /F1 11.955 Tf 34.03 0 Td[()Q33(---)Q43()C,D)Q14(|)Q24()-256()-256()]TJ /F1 11.955 Tf 34.02 0 Td[()Q34(---)Q44() ................................. 179 15
PAGE 16
6-63LongitudinalaircraftstateresponsesA)state1(u)B)state2(w)C)state3(q)D)state4() ................................... 180 6-64Longitudinalaircraftcolumneigenvectorsfor30degree,10secondmorph:A)V11(|)V21()-100()-101()]TJ /F1 11.955 Tf 30.3 0 Td[()V31(---)V41()B)V12(|)V22()-100()-101()]TJ /F1 11.955 Tf 30.3 0 Td[()V32(---)V42()C)V13(|)V23()-98()-98()]TJ /F1 11.955 Tf 30.24 0 Td[()V33(---)V43()D)V14(|)V24()-98()]TJ -391.46 -14.45 Td[()]TJ /F1 11.955 Tf 9.3 0 Td[()V34(---)V44() ............................... 181 6-65DesiredResponse(|)andOptimalResponse(\000)]TJ /F1 11.955 Tf 27.89 0 Td[()forA)First-OrderMorphingandB)Second-OrderMorphingandC)Third-OrderMorphing .......... 183 6-66DesiredResponse(|)andOptimalResponse()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()forSinusoidalMorphing 184 6-67DesiredResponse(|)andOptimalResponse()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()forSinusoidalMorphing 185 6-68A)ResponsetoMorphingTrajectory:Desired(|),Optimized()-281()-282()]TJ /F1 11.955 Tf 34.63 0 Td[()B)ResponsetoMorphingTrajectory:Desired(|),Optimized(\000)]TJ /F1 11.955 Tf 27.89 0 Td[()C)ResponsetoMorphingTrajectory:Desired(|),Optimized()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() ............ 186 6-69ResponsetoMorphingTrajectory:Desired(|)andOptimized()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[() .... 187 6-70ResponsetoMorphingTrajectory:Desired(|)andOptimized()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[() .... 187 16
PAGE 17
AbstractofDissertationPresentedtotheGraduateSchooloftheUniversityofFloridainPartialFulllmentoftheRequirementsfortheDegreeofDoctorofPhilosophyALINEARINPUT-VARYINGFRAMEWORKFORMODELINGANDCONTROLOFMORPHINGAIRCRAFTByDanielT.GrantMay2011Chair:RickLindMajor:AerospaceEngineering Morphing,whichchangestheshapeandcongurationofanaircraft,isbeingadoptedtoexpandmissioncapabilitiesofaircraft.Theintroductionofbiological-inspiredmorphingisparticularlyattractiveinthathighly-agilebirdspresentexamplesofdesiredshapesandcongurations.Apreviousstudyadoptedsuchmorphingbydesigningamultiple-jointwingthatrepresentedtheshoulderandelbowjointsofabird.Theresultingvariable-gullaircraftcouldrotatethewingsectionverticallyatthesejointstoaltertheightdynamics.Thispaperextendsthatmultiple-jointconcepttoallowavariable-sweepwingwithindependentinboardandoutboardsections.Theaircraftisdesignedandanalyzedtodemonstratetherangeofightdynamicswhichresultfromthemorphing.Inparticular,thevehicleisshowntohaveenhancedcrosswindrejectionwhichisacertainlycriticalmetricfortheurbanenvironmentsinwhichtheseaircraftareanticipatedtooperate. Missioncapabilitycanbeenabledbymorphinganaircrafttooptimizeitsaerodynamicsandassociatedightdynamicsforeachmaneuver.Suchoptimizationoftenconsiderthesteady-statebehavioroftheconguration;however,thetransientbehaviormustalsobeanalyzed.Inparticular,thetime-varyinginertiashaveaneectontheightdynamicsthatcanadverselyaectmissionperformanceifnotproperlycompensated.Theseinertiatermscausecouplingbetweenthelongitudinalandlateral-directionaldynamicsevenformaneuversaroundtrim.Asimulationofavariable-sweepaircraftundergoingasymmetric 17
PAGE 18
morphingforanaltitudechangeshowsanoticeablelateraltranslationintheightpathbecauseoftheinducedasymmetry. Theightdynamicsofmorphingaircraftmustbeanalyzedtoensureshape-changingtrajectorieshavethedesiredcharacteristics.Thetoolsfordescribingightdynamicsofxed-geometryaircraftarenotvalidfortime-varyingsystemssuchasmorphingaircraft.Thispaperintroducesamethodtorelatetheightdynamicsofmorphingaircraftbyinterpretingatime-varyingeigenvectorintermsofightmodes.Thetime-varyingeigenvectorisactuallydenedthroughadecompositionofthestate-transitionmatrixandthusdescribesanentireresponsethroughamorphingtrajectory.Avariable-sweepaircraftisanalyzedtodemonstratetheinformationthatisobtainedthroughthismethodandhowtheightdynamicsarealteredbythetime-varyingmorphing. Also,morphingvehicleshaveinherentlytime-varyingdynamicsduetothealterationoftheircongurations;consequently,thenumeroustechniquesforanalysisandcontroloftime-invariantsystemsareinappropriate.Therefore,acontrolschemeisintroducedthatdirectlyconsidersaconceptoftime-varyingpoletocommandmorphing.Theresultingtrajectoryminimizingtrackingerrorforeitherastateresponseorapoleresponse. 18
PAGE 19
CHAPTER1INTRODUCTION 1.1Motivation Miniatureairvehiclesareresourceswhosecharacteristics,suchassizeandspeed,enablearangeofmissionproles.Thesevehiclesareideallysuitedtooperatewithinurbanenvironmentsataltitudesthatareinaccessibletolargeraircraftduetodenseobstacles.Tasksincludingsurveillanceandtrackingwillbegreatlyfacilitatedbyvehiclesthatcanyattreetoplevelandintobuildings.Anillustrationofapossiblesurveillancemission,astraversedthroughanurbanenvironment,maybeseeninFig. 1-1 Figure1-1. Surveillancemissionthroughanurbanenvironment Agilityisincreasinglyrequiredforthesevehiclesasthemissiontasksconsidertheightconditionsassociatedwithurbanenvironments.Theclosespacingofobstacleswillrequireavehiclethatcanturnsharplyinasmallradiusbutyetloiterandcruise.Thewindsaroundtheseobstaclessignicantlyvaryindirectionwhichwillrequirethevehicletoincurlargeanglesofsidesliptomaintainsensorpointing.Thedurationforwhichthesensorismaintainedonthetargetiscrucialtocompletingmissionobjectivessuchaslaser-basedswathmapping[ 108 ]andvision-basedpathplanning[ 64 ],asshowninFig 1-2 19
PAGE 20
Figure1-2. Vision-basedpathplanning Suchdisparaterequirementsplaceconstraintsonthedesignwithinwhichasinglevehiclecannotlie.Therefore,morphingisbeingincorporatedtoenablemulti-rolecapabilitiesofasinglevehicle.Essentially,thevehiclechangesshapebyalteringparameters,suchasspanorcamber,duringight.Theresultingrangeofcongurationswillhaveanassociatedrangeofightdynamicsand,consequently,maneuvering. 1.2ProblemDescription BattleeldenvironmentshavepreviouslyservedasaperformancestageformanyprovencommercialgradeUAVs.TheseUAVsaretypicallylargerinsize,anddesignedprimarilyforsurveillancefromhigheraltitudes,relativetoitssmallercounterpart,theMAVorMiniatureAerialVehicle.ThelargerUAVsmaylacktheadvantageofsizeandmaneuverabilityovertheMAV,butsignicantlymake-upforthefactwiththeirtechnologicalreadinessformissioncapability,asshowninFig. 1-3 Modernavionics,andtheirrespectivesub-systems,haverecentlymadelargeadvancesinthereductionoftheiroverallweightandsize.Asaresult,MAVsarebeginningtobeoutttedwithmoresophisticatedsensorpackagesandcontrolsystems.Itshouldbenotedhowever,thatevenwiththeseadvances,thelargerUAVisstillsuperiortotheMAVintermsofbeingmissioncapable.Duetothefactofthistechnologicalgap,ithasbeensuggestedthataviancharacteristics,andtheireectivebenets,bestudied. 20
PAGE 21
Figure1-3. Readinessformissioncapability TwoMAVsutilizingtheconceptofwingmorphing,orshapechanging,(acommoncharacteristicofavianight)canbeshowninFig. 1-4 A B Figure1-4. MorphingMAVs:A)CapableofhorizontalmorphingB)Capableofverticalmorphing TheightdynamicsoftheaircraftshowninFig. 1-4 aresomewhatuniqueanddierentfromthosedenedforsymmetricxed-wingight.Withtheintroductionofmorphing,afewpreviouslymadeassumptionsmustnowbereconsidered.Itshouldbenotedthatmorphingchangesthesystemfromtime-invarianttotime-varying,andasaresult,introducesnewinertialtermsintothedynamics. 21
PAGE 22
Themomentsofinertiaofabodyobviouslyhaveaprofoundinuenceonthedynamicsandassociatedmotionofthatbody.Certainlyaerospacesystems,withmultipledegreesoffreedomfortranslationandrotation,mustproperlyaccountforinertiatoahighlevelofaccuracyinordertomodelthedynamics.Thetime-varyingaspectoftheseinertiasmustbeconsideredwithsimilaraccuracytonoteitsinuenceonthesystem. Someextensiveandrigorousevaluationsoftraditionalcausesoftime-varyinginertia,suchasfuelexpenditureandmulti-bodyrotation,havebeenperformedforspacesystems.Theeectsoftranslatingmasswithinaspacestationarederivedunderanassumptionofharmonicmotionandusedtocomputelibrationalstability[ 114 ].Movingmasswasalsoincludedinthedynamicsofavehiclewithasolarsailthatcouldmoveforcontrolpurposes[ 124 ].Thedynamicsandassociatedtime-varyinginertiawasmodeledforatwo-vehicleformationinwhichaCoulombtethercontrolledtherelativedistanceandmassdistribution[ 82 ].Anotherstudyoptimizedadesignforatwo-vehicleformationwithaexibleappendagewhosemotionalteredtheinertiaproperties[ 89 ].Theinuenceofthrusters,whichexpendmassthroughactivationandthusvarytheinertia,wasinvestigatedusingaformulationoffeedbackandfeedforwardtocanceltheeects[ 118 ].Thetime-varyinginertiaduetothrusterswascoupledwitheectsofuidsloshinginanotherexaminationofspacecraftdynamics[ 51 ]. Thesetraditionalcauseshavealsobeenexaminedwithrespecttotheireectsonaircraftalthoughnotnecessarilytothesamedegreeasspacecraft.Fuelburnisoftenneglectedsinceitstimeconstantisslowerthantheightdynamicsofmanyaircraft;however,thateectsontime-varyinginertiawereshownforthecaseofaerialrefuelinginwhichmasswasrapidlytransferredfromthetankertotherecipient[ 120 ].Onasmallerscale,thedynamicsofaapping-wingmicroairvehiclewerestudiedbynotingtheeectofwingmotion[ 121 ]. Theintroductionofmorphing,orshape-changingactuation,toanaircraftwillaltertheshapeandmassdistributionofthevehicle,andasaresult,producetime-varying 22
PAGE 23
inertias.Manystudiesintomorphingaircrafthavefocusedonthesteady-statebenetsofalteringacongurationforissuessuchasfuelconsumption[ 14 ],rangeandendurance[ 43 ],costandlogistics[ 15 ],actuatorenergy[ 93 ],maneuverability[ 99 ]andairfoilrequirements[ 103 ].Additionally,aeroelasticeectshavebeenoftenstudiedrelativetomaximumrollrate[ 66 44 9 ]andactuatorloads[ 77 ]. Morphinghasalsobeenintroducedtomicroairvehiclesforthepurposeofmanueveringcontrol[ 1 2 ].Specically,anaircraftisdesignedthatusesindependentwing-sweep,asshowninFig. 1-4 ,ofinboardandoutboardsectionsonboththerightandleftwings[ 46 ].Thataircraftisshowntousethemorphingforalteringtheaerodynamicsandachieveperformancemetricsrelatedtosensorpointing.Thewingsareabletosweepontheorderofasecond;consequently,thetemporalnatureofthemorphingmustbeconsidered. 1.3ProblemStatement Performanceofamorphingaircraftiscriticaltothesuccessofanymission;therefore,methodstoachievethisperformancemustbeconsidered.Designinganeectivecontrolschemeformorphingaircraftisanon-trivialtask,butcanbemademoreapparentbyrstunderstandingthesystemdynamics.Thedynamicperformanceoflinearsystemshaveclassicallybeenstudiedusingtime-invariantmethodssuchaseigenvalueandeigenvectoranalysis[ 97 ].Thisanalysishelpsdeterminesystemstabilityaswellasdynamicightmodes. Morphingcanbemodeledasatime-invariantsystemwithdiscretechanges[ 46 ],butisconsideredmoreaccuratewhenmodeledasatime-varyingsystem.Itiswellknown[ 127 ]thattime-invarianteigenvaluesprovidenoinformationabouttime-varyingstability.Asaresult,conceptsanalogoustotime-invarianteigenvaluesandeigenvectorshavebeendevelopedfortime-varyingsystems[ 59 86 126 130 131 127 ].Theparticularmethodologyforwhichtheseconceptsarederived,varybaseduponthesystemrepresentation.Oneapproachaddressesthenotionofpolesetsderivedfromadierentialequationwithtime-varyingcoecients[ 59 131 ].Anotherapproachaddressesthenotionofpolesets 23
PAGE 24
derivedbytransformingthestateequationintoanupper-triangularstateequation,viaastabilitypreservingvariablechange[ 86 ]. Thechallengeariseswhenconsideringmorphingasaplausiblecontrolscheme.Typically,anaircraftiscontrolledaboutitsaxesbycommandingadeection,fromrelativetrim,inoneofitsclassicallydenedcontrolsurfaces.Itisseenhowever,thatnocontrolsurfacesarepresentinavianight,yetmaneuveringisstilldoneeciently.Thisobservationcomesasadirectresultofwingmorphing,andthus,servesasaeldofinterestincontroldesign.Themainthoughtbeing,isitpossibletoeectivelycontrolanaircraftwithmorphing,andifitis,howdoyoudescribetheresultingtime-varyingeects? 1.3.1Contributions TheclassofsystemsunderinvestigationcanberepresentedasinEquation 1{1 .Thisformdescribesasystemwithnxstatesalongwithnmorphingactuatorsandnutraditionalactuatorssuchthatx2Rnxisthestatevector,2Rnisthecommandtomorphing,andu2Rnuisthecommandtotraditionalcontrolsurfaces.ThecoecientsintheequationsofmotionarecombinedintoA2RnxnxandB2Rnxnuwhicharefunctionsofboththemorphingconguration,,andtheightcondition,. _x=A(;)x+B(;)u(1{1) Thisresearchhasresultedinthedevelopment,building,andtestingofinnovativeandtheoreticaladvancementsusedtoaddressthechallengesrelatedtotime-varyingmorphingaircraft.Asetofsolutionsareoutlinedinthedissertationtoaddressthisnovelapproachdevelopedforsystemsanalysisandcontrolsynthesis.Theresultingapproachhasresultedinthefollowingspeciccontributions. MorphingAircraft { Amorphingplatformwasdesignedsuchthatactuationofanindependentmulti-joint,asymmetricwingwasmadepossible. { Missionperformancewasevaluatedforthisaircraftthroughsimulationandighttesting. 24
PAGE 25
Time-VaryingInertias { Retainedallinertialtermsintheequationsofmotiontostudytheirtime-varyingeectswithinmorphingightdynamics. { Theresultingtime-varyinginertialeectswerecharacterizedandparamterizedbasedontheirinuencewithinthemodalstructureoftheightdynamics. Time-varyingFlightDynamics { Interpretationsweredevelopedforthetime-varyingpolesandmodalcharacteristicsofamorphingaircraft. { Theseinterpretationswerecompletedusingtwodierentmethodologiespreviouslyuntestedfortime-varyingightdynamics. MorphingControl { Aframeworkforlinearinput-varyingsystemswasformulatedtoaccountforavarietyofsystemswhosedynamicsaredependentinasetofcontroleectors. { Developedandsynthesizedoptimalfeedforwardtrackingfordesiredtime-varyingpolesandsignals. { Developedandsynthesizedoptimalfeedbackforaspeciccaseoftime-varyingsystem.Thissystemwasredenedinabilineartime-invariantformwherenoinputmatrixwasavailableforcontrolpurposes. 1.3.2Papers ThecontributionslistedinSection 1.3.1 haveresultedinmultiplepublicationsincludingjournalpapers,conferencepapers,andabookchapter,aswellas,televisiondocumentariesandaU.S.patent.Thesepublications,alongwiththedocumentariesandpatent,arelistedasfollows: BookChapter 1. D.T.Grant,S.Sorley,A.Chakravarthy,R.Lind,"FlightDynamicsofMorphingAircraftwithTime-VaryingInertias,"inMorphingVehiclesandStructures:AnAerospacePerspective,editedbyJ.Valasek,Wiley[Acceptedforpublication,scheduledforreleaseJanuary2011] JournalPapers 25
PAGE 26
2. D.T.Grant,M.AbdulrahimandR.Lind,"DesignandAnalysisofBiomimeticJointsforMorphingofMicroAirVehicles,"InternationalJournalofMicroAirVehicles,[Accepted]. 3. D.T.Grant,M.AbdulrahimandR.Lind,"FlightDynamicsofaMorphingAircraftUtilizingIndependentMultiple-JointWingSweep,"BioinspirationandBiomimetics,[Accepted]. 4. A.Chakravarthy,D.T.GrantandR.Lind,"Time-VaryingDynamicsofMicroAirVehicleWithVariable-SweepMorphing,"JournalofGuidance,Control,andDynamics,[InReview]. 5. D.T.GrantandR.Lind,"OptimalFeedbackControlforTime-VaryingMorphingAircraft,"[FutureWorkinProgress]. ConferencePapers 6. D.T.Grant,M.AbdulrahimandR.Lind,"FlightDynamicsofaMorphingAircraftUtilizingIndependentMultiple-JointWingSweep,"AIAAAtmosphericFlightMechanicsConference,Keystone,CO,August2006,AIAA-2006-6505.[BestStudentPaper] 7. D.T.Grant,M.AbdulrahimandR.Lind,"EnhancingMissionCapabilityforMicroAirVehiclesusingBiomimeticJointsandStructures,"AustralianInternationalAerospaceCongress,Melbourne,Australia,March2007,accepted. 8. D.T.Grant,M.AbdulrahimandR.Lind,"TheRoleofMorphingtoEnhanceMissionCapabilityofMicroAirVehicles,"WorldForumonSmartMaterialsandSmartStructuresTechnology,invitedpaper,ChongqingandNanjing,China,May2007. 9. D.T.Grant,M.AbdulrahimandR.Lind,"AnInvestigationofBiologically-InspiredJointstoEnableSensorPointingofMorphingMicroAirVehicles,"Interna-tionalForumonAeroelasticityandStructuralDynamics,Stockholm,Sweden,June2007,IF-069. 10. D.T.GrantandR.Lind,"EectsofTime-VaryingInertiasonFlightDynamicsofanAsymmetricVariable-SweepMorphingAircraft,"AIAAAtmosphericFlightMechanicsConference,HiltonHead,SC,August2007,AIAA-2007-6487. 11. D.T.Grant,A.ChakravarthyandR.Lind,"ModalInterpretationofTime-VaryingEigenvectorsofMorphingAircraft,"AIAAAtmosphericFlightMechanicsCon-ference,Chicago,IL,AIAA-2009-5848. 12. D.T.GrantandR.Lind,"OptimalTrackingofTime-VaryingModesForControlofMorphingAircraft,"AIAAGuidance,Navigation,andControlConference,Toronto,ON,Canada,August2010,AIAA-2010-8203. 26
PAGE 27
13. A.Chakravarthy,D.T.GrantandR.Lind,"Time-VaryingDynamicsofaMicroAirVehiclewithVariable-SweepMorphing,"AIAAGuidance,NavigationandControlConference,Chicago,IL,August2009,AIAA-2009-6304. TelevisionDocumentariesandFilms 14. "FlyingMosters3D":Sky3DTVDocumentary(3DImaxFeatureFilm)[AiredDecember2010] 15. "MorphingMAV":NationalGeographicChannel(5minutesegmentonMadLab)[FirstAired:Apr2008] 16. "BirdWingsforAircraft":VPRO-publictelevisioninTheNetherlands(25minutesegmentonCopyCats)[FirstAired:Dec2006] 17. "TheMagicofMotion":PBS(20minutesegmentonNatureTechDocumentary)[FirstAired:Nov2006] 18. "MorphingMAV":ScienceChannel(5minutesegmentonDiscoveriesThisWeek)[FirstAired:Jul2006] 19. "Biologically-InspiredMorphingMAV":DiscoveryChannel(5minutesegmentonBeyondTomorrow)[FirstAired:May2006] 20. "MorphingMAV":DiscoveryChannel(5minutesegmentonDailyPlanet)[FirstAired:Oct2005] Patent 21. MorphingAircraft:USPatentApplicationSerialNumberPCT/US09/54917 Thepublicationslistedabovearederivedfromcertainsectionsofthisdissertation.Relationshipsbetweenthesetopics,dissertationsections,andresultingpublicationscanbefoundinTable 1-1 Table1-1. Publicationsresultingfromresearch TopicDissertationSectionWork PlatformDesign 6.1 6.3 1. 2. 3. 6. 7. 8. 9. Time-VaryingInertias 2.1 2.6 6.4 10. Time-VaryingFlightDynamics 3.1 3.6 4.1 4.4 4. 11. 13. OptimalFeedforwardControl 5.1 5.4 12. OptimalFeedbackControl 5.1 5. 27
PAGE 28
1.4DissertationOverview Chapter1introducesthefollowingdocumentbystrategicallyaddressingthekeyissuesdirectlyrelatedtothedissertation'smaintopicofresearch.Theseissuesarewrittenintosectionsbasedupontheircontributiontotheoverallunderstandingoftheproblembeingconsidered.Sectiontopicsinclude:motivationforconductingresearch,abackgrounddescriptionoftheproblem,andastatementformallydeningtheactualproblem. Chapter2beginsbyrecallingthetraditionalderivationandlinearizationofthenonlinearaircraftequationsofmotion.Theseaircraftequationsofmotionarethenlaterrederived,andlinearized,basedupontheassumptionthatthesystemistime-varying(aresultofmorphing).Thischapterisconcludedbyexaminingtwoexampleswhereboththesymmetricandasymmetricmorphingequationsofmotionarereducedtothetraditionallydenedequationsofmotion,asdescribedpreviouslyinthechapter. Chapter3updatesthereaderwiththemostinuentialmethodsindeningthepolesandzerosforalineartime-varyingsystem.Thefourmost-commonapproachesandtheirmethodsaregiven,eachincludingabriefdescriptionofbackgroundtheoryanddenitions.Thestabilityofeachmethodisalsoconsidered,therefore,necessaryandsucientconditionspertainingtodierentlevelsofstabilityaredened. Chapter4beginsbyreviewingthefundamentalframeworksusedtointerpretbothtime-invariantandtime-varyingsystemcharacterisitcs.Theresultingtime-varyingframeworkisthenfurtherinvestigatedfortwoofthefourmethodologiesdescribedinChapter3.Chapter4closesbyprovidingatime-varyingexamplespecictotheframeworkpreviouslymention. Chapter5motivesthederivationofasystematicapproachtoclassifyingandcontrollingthetime-varyingsystemproducedbymorphing.Morphingcontrolchallangesandapproachesaredecribedandthenfollowedbyexamplesoutliningcertainapproaches.Bothstabilizingandoptimizingperformanceareconsideredthroughtheoreticalderivationandthenimplementation.Thechatpterconcludeswithafutureworksectionwhichexpandsthepreviousworkintopossibleavenuesforfutureworkandproposestheoreticalderivations. Chapter6presentsanextensiveexamplewhichinvestigatestheeectsofbothsymmetricandasymmetricmorphing.Generatedstate-spacesystemsareusedtosimulatebasiclineartime-invariant(quasi-static)controlledmaneuverssuchasdivingandturning,aswellas,theeectsoftime-varyingmorphingonthesystem'smodalcharacterisiticsandcontrollability. 28
PAGE 29
CHAPTER2EQUATIONSOFMOTION 2.1AircraftAxisSystem Threeaxissystemsarecommontodescribeaircraftmotion.Thesesystemsincludethebody-axissystem(xedtotheaircraft),theEarth-axissystem(assumedtobeaninertialaxissystemxedtotheEarth),andthestability-axissystem(denedwithrespecttothelocalwind). 2.1.1BodyAxisSystem Thebody-xedcoordinatesystemhasitsoriginlocatedattheaircraft'scenterofgravity.Theaxesareorientedsuchthat^xB,pointsdirectlyoutthenose,^yB,pointsdirectlyouttherightwing,and^zB,pointsdirectlyoutthebottom,asshowninFigure 2-1 Figure2-1. Body-FixedCoordinateFrame 2.1.2StabilityAxisSystem Thestabilityaxissystemsharesthesameoriginlocationasthebody-xedframe,butisrotatedrelativetothebody-xedaxistoalignwiththevelocityvector,asshowninFigure 2-2 .Theresulting^xSaxispointsinthedirectionoftheprojectionoftherelativewindontothexzplaneoftheaircraft.The^ySaxisisoutoftherightwingcoincidentwith^yB,whilethe^zSaxispointsdownwardinthedirectioncompletingthevectorsetdescribedbytheright-handruleandshowninFigure 2-2 29
PAGE 30
Figure2-2. StabilityCoordinateFrame 2.1.3EarthAxisSystem TheEarth-xedcoordinatesystemisxedtothesurfaceoftheEarthwithits^zEaxispointingtothecenteroftheEarth.The^xEand^yEaxesareorthogonalandlieinthelocalhorizontalplane,asshowninFigure 2-3 Figure2-3. Earth-FixedCoordinateFrame 30
PAGE 31
2.2CoordinateTransformations 2.2.1EarthtoBodyFrame AvectormaybetransformedfromtheEarth-xedframeintothebody-xedframebythreeconsectutiveconsecutiverotationsaboutthez-axis,y-axis,andx-axis,respectively.TraditionalightmechanicsdenetheanglesthroughwhichthesecoordinateframesarerelativelyrotatedastheEulerangles.TheEuleranglesareexpressedasyaw ,pitch,androll,anditisimportanttonotethataparticularsequenceofEuleranglerotationsisunique.Thefollowingillustratesthetransformationofavector,PE,intheEarth-xedframe,asdenedinEquation 2{1 ,intothebody-xedframe. PE=x^iE+y^jE+z^kE=266664XEYEZE377775(2{1) Therstrotationisthroughtheangleaboutthevector^kE,asshowninFigure 2-4 Figure2-4. Rotationthrough Therotationabout^kEbytheangle, ,isreferredtoasR3( ),whereR3( )istheshort-handnotationusedtodescribetherotationmatrixdenedinEquation 2{2 31
PAGE 32
266664X1Y1Z1377775=266664cos sin 0)]TJ /F1 11.955 Tf 11.3 0 Td[(sin cos 0001377775266664XEYEZE377775(2{2) Thesecondrotationisthroughtheangleaboutthevector^j1=^j2,asshowninFigure 2-5 Figure2-5. Rotationthrough Therotationabout^j1bytheangle,,isreferredtoasR2(),whereR2()istheshort-handnotationusedtodescribetherotationmatrixdenedinEquation 2{3 266664X2Y2Z2377775=266664cos0)]TJ /F1 11.955 Tf 11.29 0 Td[(sin010sin0cos377775266664X1Y1Z1377775(2{3) Thethirdandnalrotationisthroughtheangleaboutthevector^i2=^i,asshowninFigure 2-6 Therotationabout^i2bytheangle,,isreferredtoasR1(),whereR1()istheshort-handnotationusedtodescribetherotationmatrixdenedinEquation 2{4 32
PAGE 33
Figure2-6. Rotationthrough 266664XBYBZB377775=2666641000cossin0)]TJ /F1 11.955 Tf 11.3 0 Td[(sincos377775266664X2Y2Z2377775(2{4) Therefore,anyvectorintheEarth-xedframe,PE,canbetransformedintothebody-xedframe,PB,usingtherelationshipdenedinEquation 2{5 PB=R1()R2()R3( )PE(2{5) Forexample,Earth-xedgravityforcescanbeexpressedinthebody-xedcoordinatesystembyimplementingEquation 2{5 ontheEarth-xedweightvector FGE=26666400mg377775E=266664)]TJ /F5 11.955 Tf 9.3 0 Td[(mgsinmgsincosmgcossin377775B(2{6) 2.2.2StabilitytoBodyFrame Typically,aerodynamicforcesaredenedbythestabilityaxisforconvenienceyet,forcomputationalreasons,theyneedtobedenedbythebody-xedaxis.Recallingthatthistransformationcanbeaccomplishedbyrotatingthestabilityaxisthroughapositiveangleofattack,thetransformationfromthestabilitycoordinateframetothebody-xed 33
PAGE 34
coordinateframeissimplyarotationaboutthevector^yBthroughtheangle,asdenedinEquation 2{7 266664FAxFAyFAz377775B=266664cos0)]TJ /F1 11.955 Tf 11.29 0 Td[(sin010sin0cos377775266664)]TJ /F5 11.955 Tf 9.29 0 Td[(DFAy)]TJ /F5 11.955 Tf 9.29 0 Td[(L377775S(2{7) 2.3NonlinearEquationsofMotion 2.3.1DynamicEquations TherigidbodyequationsofmotionareobtainedformNewton'ssecondlaw,whichstatesthatthesummationofallexternalforcesactingonabodyisequaltothetimerateofchangeofthemomentumofthebody;andthesummationoftheexternalmomentsactingonthebodyisequaltothetimerateofchangeofthemomentofmomentum(angularmomentum).ThetimeratesofchangeoflinearandangularmomentumarerelativetoaninertialorNewtonianreferenceframe(Earth-xedframe).Newton'ssecondlawcanbeexpressedbythevectorsdenedinEquations 2{8 2{9 XF=d dt(mv)E(2{8) XM=d dtHE(2{9) 2.3.1.1Forceequations Afewadditionalassumptionsmustbemadeinordertodeveloptheright-handside,orresponseside,ofequation 2{8 .Theseassumptionsincludethattheaircraftbeconsideredarigidbody,andthatthemassoftheaircraftremainconstant.Asaresultofassumingthemasstobeconstant,themassterm,seeninEquation 2{8 ,canbemovedoutsideofthetimederivativeandredenedasEquation 2{10 34
PAGE 35
XF=md dtvE=maE(2{10) Theaccelerationofanaircraftisnormallymeasuredinthebody-xedframe,therefore,touseEquation 2{10 ,thebody-xedaccelerationmustbetransformedintotheEarth-xedframe.Itisnotedthatsincethetransformationinvolvesarotationofcoordinates,anyvectorinthebody-xedframecanbecalculatedintheEarth-xedframebyusingthetransporttheorem[ 95 ],asseeninEquation 2{11 .Itshouldalsobenotedthatforthefollowingderivations,allofthevectorswillbeexpressedinthebody-xedcoordinatesystem,denotedbythesubscript,B,unlessotherwisestated.Forexample,thevelocityvectorasseenbyanobserverintheEarth-xedframe,expressedinthebody-xedcoordinatesystem,willbedontedasvEB. db dtA=db dtB+A!Bb(2{11) Thevelocityvector,vE,representstherateofchangeofthepositionvector,r,asviewedbyanobserverintheEarthxedreferenceframe.Sincethepositionvector,r,isnormallymeasuredwithrespecttothebody-xedcoordinatesystem,itmustbersttransformedintotheEarth-xedframe,inordertocomputethedesiredEarth-denedvelocityvector,VEB.Thus,thetransporttheoremisusedtorelatethepositionvector,r,totheEarth-denedvelocityvector,vEB,asseeninasseeninEquation 2{12 vEB=dr dtE=dr dtB+E!Br(2{12) TheEarth-denedveloctyvectorcanbeconvenientlyrewrittenanddened,asseeninEquation 2{13 35
PAGE 36
vEB=u^i+v^j+w^k=266664UVW377775EB(2{13) ThesttermontherightsideofEquation 2{12 iscalledthebody-denedvelocity.Thebody-denedvelocityvector,vBB,issimplythetimederivativeofthepositionvector,r,asseeninEquation 2{14 .Notethatthisisonlythecasewhenthepositionvectorisdescribedbythebody-xedcoordinatesystem. dr dtB=_x^i+y^j+_z^k=vBB(2{14) ThesecondtermontherightsideofEquation 2{12 ,isdescribedbythecrossproductbetweentheEarth-denedvelocityvector,vEBandatermcalledtheangularvelocityvector.Theangularvelocityvector,E!B,describestheangularvelocityofreferenceframeB(body-xedframe)asviewedbyanobserverinreferenceframeE(Earthxedframe),representedinthebody-xedcoordinatesystem. Theangularveloctyvectorcanbeconvenientlyrewrittenanddened,asseeninEquation 2{15 B!E=p^i+q^j+r^k=266664_p_q_r377775EB(2{15) Itshouldbenoted,thattheangularvelocityvectors,E!BandB!E,arerelatedthroughasimplerelationship,asseeninEquation 2{16 E!B=)]TJ /F11 7.97 Tf 9.3 4.94 Td[(B!E(2{16) Itshouldalsobenoted,thatangularvelocityvectorsrelatingmultiplerotations,maybelinearlyaddedintooneangularvelocityvectorthatrelatestheentiretransformation. 36
PAGE 37
Thisadditionofangularvelocityvectorscanbedescribedbytheangularvelocityadditiontheorem[ 95 ],whichisshowninEquation 2{17 A1!An=A1!A2+A2!A3+:::+An)]TJ /F15 5.978 Tf 5.75 0 Td[(1!An(2{17) ThederivationoftheEarth-denedaccelerationvector,aEB,issimilartothepreviousdenitionoftheEarth-denedvelocityvector,vEB.Thatis,theEarth-denedaccelerationvector,aEB,representstherateofchangeoftheEarth-denedvelocityvector,vEB,asviewedbyanobserverintheEarth-xedreferenceframe.Therefore,tosolvefortheaccelerationvector,aEB,thetransporttheoremmustbeappliedtothevelocityvector,vEB,asseeninEquation 2{18 aEB=dvEB dtE=dvEB dtB+E!BvEB(2{18) TheEarth-denedaccelerationvectorcanbeconvenientlyrewrittenanddened,asseeninEquation 2{19 aEB=u^i+v^j+w^k=266664_u_v_w377775B(2{19) NoticethatifthesttermontherightsideofEquation 2{19 isdenedbythebody-xedcoordinatesystem,thenonlythetimederivativeneedstobetaken.Ifthistermisnotdenedbythebody-xedcoordinatesystem,thenthetransporttheoremmustbeappliedtocompensateforthecoordinatechange. TherighthandsideofEquation 2{19 cannowbesolvedforbyrstcomputingthecrossproductbetweentheEarth-denedangularvelocityvector,B!E,andtheEarth-denedvelocityvector,vEB.Thecrossproductisthenaddedtothebody-denedaccelerationvector,aBB.Itshouldnotedthatthebody-denedaccelerationisnotsimplythesecondderivativeoftheofthebody-denedposition.Thistermisdenedastherate 37
PAGE 38
ofchangeoftheEarth-denedvelocityvector,asseenbyanobserverinthebody-xedframe,andtherefore,canbecomputedbytakingthetimederivativeoftheEarth-denedvelocityvector. Theresultingterm,showninEquation 2{20 ,representstheaccelerationoftheaircraftasseenbyanobserverintheinertially-xedEarthframe,representedinthebody-xedcoordinatesystem. aB=266664_u+qw)]TJ /F5 11.955 Tf 11.96 0 Td[(rv_v+ru)]TJ /F5 11.955 Tf 11.95 0 Td[(pw_w+pv)]TJ /F5 11.955 Tf 11.95 0 Td[(qu377775B(2{20) Todenethelefthandside,orappliedforceside,ofEquation 2{8 ,itisrstassumedthatonlythemostsignicantforcesaectthemotionoftheaircraft.Theappliedforcescanthenbebrokendownintovectorcomponentsandarrangedinamannersuchthattheyaredenedbythevector,FEB,asseeninEquation 2{21 FEB=266664FxFyFz377775EB=266664FGx+FAxFGy+FAyFGz+FAz377775EB=XF(2{21) Aresultingsetoffull-order,nonlinearforceequations,asseeninEquation 2{22 ,canbederivedbyinsertingEquations 2{20 2{21 intoEquation 2{8 andrecallingthattheappliedforcesaregivenbyEquations 2{6 2{7 m(_u+qw)]TJ /F5 11.955 Tf 11.96 0 Td[(rv)=)]TJ /F5 11.955 Tf 9.3 0 Td[(mgsin+()]TJ /F5 11.955 Tf 9.3 0 Td[(Dcos+Lsin)m(_v+ru)]TJ /F5 11.955 Tf 11.96 0 Td[(pw)=mgsincos+FAym(_w+pv)]TJ /F5 11.955 Tf 11.96 0 Td[(qu)=mgcoscos+()]TJ /F5 11.955 Tf 9.29 0 Td[(Dsin)]TJ /F5 11.955 Tf 11.96 0 Td[(Lcos)(2{22) 38
PAGE 39
2.3.1.2Momentequations Itisshowninequation 2{2 thatNewton'ssecondlawissatisedbyequatingthesummationofmomentstothetotalrateofchangeofthemomentofmomentum(angularmomentum).ThesamerelationshipusedpreviouslytodenetheEarth-denedaccelerationcanbeusedtodenetheEarth-denedangularmomentum.Itshouldbeagainnotedthatforthefollowingderivations,allofthevectorswillbeexpressedinthebody-xedcoordinatesystem,denotedbythesubscript,B,unlessotherwisestated.Forexample,theangularmomentumvectorasseenbyanobserverintheEarth-xedframe,expressedinthebody-xedcoordinatesystem,willbedontedasHEB. Traditionally,theangularmomentumiscomputedfrommeasurementstakeninthebody-xedcoordinatesysytem.InordertoproperlydescribetheEarth-denedangularmomentumvector,itmustbersttransformedintotheinertially-xedEarthframe.Thistransformationcanbeaccomplishedbythetransporttheorem,asseeninEquation 2{23 dHEB dtE=dHEB dtB+E!BHEB(2{23) Thegeneralexpressionforangularmomentumcanbetakendirectlyfrombasicphysicsanddescribedasanobject'sinertialtensor,I,multipliedbythatobject'sangularratevector,!,asseeninEquation 2{24 H=I!(2{24) TheangularmomentumdenedinEquation 2{24 canbemaderelavanttoanaircraftbydeningtheinertialtensorintheaircraft'sbody-xedcoordinatesystem,asseeninEquation 2{25 ,andrecallingthattheEarth-denedangularvelocityvectorwaspreviouslydenedinEquation 2{15 39
PAGE 40
IB=266664Ixx)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixy)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixz)]TJ /F5 11.955 Tf 9.3 0 Td[(IyxIyy)]TJ /F5 11.955 Tf 9.3 0 Td[(Iyz)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixz)]TJ /F5 11.955 Tf 9.3 0 Td[(IyzIzz377775B(2{25) Equations 2{15 and 2{25 canbeinsertedintoEquation 2{24 toproducetheEarth-denedangularmomentumvector,asseeninEquation 2{26 HEB=266664Ixx)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixy)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixz)]TJ /F5 11.955 Tf 9.3 0 Td[(IyxIyy)]TJ /F5 11.955 Tf 9.3 0 Td[(Iyz)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixz)]TJ /F5 11.955 Tf 9.3 0 Td[(IyzIzz377775EB266664PQR377775EB(2{26) TheEarth-denedangularmomentumvector,HEB,canbeconvenientlywritteninvectornotation,asseeninEquation 2{27 HEB=Hx^i+Hy^j+Hz^k=266664HxHyHz377775EB(2{27) InsertingEquation 2{27 intothelefthandsideofEquation 2{26 andcarryingoutthematrixmultiplicationontherighthandside,resultsintheEarth-denedvectornotationoftheaircraft'sangularmomentum,asseeninEquation 2{28 HEB=266664pIx)]TJ /F5 11.955 Tf 11.95 0 Td[(qIxy)]TJ /F5 11.955 Tf 11.96 0 Td[(rIxzqIy)]TJ /F5 11.955 Tf 11.95 0 Td[(rIz)]TJ /F5 11.955 Tf 11.95 0 Td[(pIxyrIz)]TJ /F5 11.955 Tf 11.96 0 Td[(pIxz)]TJ /F5 11.955 Tf 11.95 0 Td[(qIyz377775(2{28) Themomentsofineritaaredescibedasindicatorstotheresistancetorotationaboutthataxis,asdenedinEquations 2{29 2{31 .Therefore,Ixindicatestheresistancetorotationaboutthex-axis(relativelydened).Theproductsofinertiaaredescribedasindicatorstothesymmetryoftheaircraft,asdenedinEquations 2{32 2{34 40
PAGE 41
Ix=Z(y2+z2)dm (2{29) Iy=Z(x2+z2)dm (2{30) Iz=Z(x2+y2)dm (2{31) Ixy=Z(xy)dm (2{32) Ixz=Z(xz)dm (2{33) Iyz=Z(yz)dm (2{34) Duetothefactthatthemasseswereassumedtobepointmasses,theintegralsinEquations 2{29 2{34 canbereducedto: Ix=m(y2+z2) (2{35) Iy=m(x2+z2) (2{36) Iz=m(x2+y2) (2{37) Ixy=m(xy) (2{38) Ixz=m(xz) (2{39) Iyz=m(yz) (2{40) ThecorrespondinginertialratescanbecalculatedbytakingthetimederivativeofEquations 2{35 2{40 ,asshowninEquationrefeqrates. 41
PAGE 42
_Ix=m[(2y)(_y)+(2z)(_z)]_Iy=m[(2x)(_x)+(2z)(_z)]_Iz=m[(2x)(_x)+(2y)(_y)]_Ixy=)]TJ /F5 11.955 Tf 9.29 0 Td[(m[(x)(_y)+(y)(_x)]_Ixz=)]TJ /F5 11.955 Tf 9.3 0 Td[(m[(x)(_z)+(z)(_x)]_Iyz=)]TJ /F5 11.955 Tf 9.3 0 Td[(m[(y)(_z)+(z)(_y)](2{41) Thersttermontherighthandsideofequation 2{23 isdescribedastherateofchangeoftheEarth-denedangularmomentumvectorasseenbyanoberverinthebody-xedcoordinatesystem,representedinthebody-xedcoordinatesystem.ThistermcanbefoundbysimplytakingthetimederivativeofEquation 2{28 ,asseeninEquation 2{42 dHEB dt=266664_pIx)]TJ /F1 11.955 Tf 14.12 0 Td[(_qIxy)]TJ /F1 11.955 Tf 13.78 0 Td[(_rIxz+p_Ix)]TJ /F5 11.955 Tf 11.96 0 Td[(q_Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(r_Ixz_qIy)]TJ /F1 11.955 Tf 13.78 0 Td[(_rIyz)]TJ /F1 11.955 Tf 14.24 0 Td[(_pIxy+q_Iy)]TJ /F5 11.955 Tf 11.95 0 Td[(r_Iz)]TJ /F5 11.955 Tf 11.96 0 Td[(p_Ixy_rIz)]TJ /F1 11.955 Tf 14.24 0 Td[(_pIxz)]TJ /F1 11.955 Tf 14.11 0 Td[(_qIyz+_rIz)]TJ /F5 11.955 Tf 11.96 0 Td[(p_Ixz)]TJ /F5 11.955 Tf 11.95 0 Td[(q_Iyz377775(2{42) ThesecondtermontherightsideofEquation 2{23 canbefoundbytakingthecrossproductbetweenthepreviouslydenedangularmomentumvector,HEB,asseeninEquation 2{28 ,andtheangularvelocityvector,!EB,asseeninEquation 2{19 .Thistermcanbethentemporarilydenedas,HT,andrewrittenintheformshownbyEquation 2{43 HT=266664qrIz)]TJ /F5 11.955 Tf 11.95 0 Td[(qpIxz)]TJ /F5 11.955 Tf 11.95 0 Td[(q2Iyz)]TJ /F5 11.955 Tf 11.96 0 Td[(qrIy+r2Iyz+rpIxyrpIx)]TJ /F5 11.955 Tf 11.95 0 Td[(qrIxy)]TJ /F5 11.955 Tf 11.95 0 Td[(r2Ixz)]TJ /F5 11.955 Tf 11.96 0 Td[(rpIz+p2Ixz+qpIyzpqIy)]TJ /F5 11.955 Tf 11.96 0 Td[(rpIyz)]TJ /F5 11.955 Tf 11.96 0 Td[(p2Ixy)]TJ /F5 11.955 Tf 11.95 0 Td[(pqIx+q2Ixy+rqIxz377775(2{43) TherightsideofEquation 2{23 cannowbesolvedforintermsoftheEarth-denedangularmomentumvector,HEB,byinsertingEquations 2{42 and 2{43 intoEquation 2{23 ,asseeninEquation 2{44 42
PAGE 43
HEB=266664_pIx)]TJ /F1 11.955 Tf 14.11 0 Td[(_qIxy)]TJ /F1 11.955 Tf 13.78 0 Td[(_rIxz+p_Ix)]TJ /F5 11.955 Tf 11.95 0 Td[(q_Ixy)]TJ /F5 11.955 Tf 11.95 0 Td[(r_Ixz_qIy)]TJ /F1 11.955 Tf 13.78 0 Td[(_rIyz)]TJ /F1 11.955 Tf 14.24 0 Td[(_pIxy+q_Iy)]TJ /F5 11.955 Tf 11.96 0 Td[(r_Iz)]TJ /F5 11.955 Tf 11.96 0 Td[(p_Ixy_rIz)]TJ /F1 11.955 Tf 14.24 0 Td[(_pIxz)]TJ /F1 11.955 Tf 14.11 0 Td[(_qIyz+_rIz)]TJ /F5 11.955 Tf 11.96 0 Td[(p_Ixz)]TJ /F5 11.955 Tf 11.96 0 Td[(q_Iyz377775+266664qrIz)]TJ /F5 11.955 Tf 11.96 0 Td[(qpIxz)]TJ /F5 11.955 Tf 11.96 0 Td[(q2Iyz)]TJ /F5 11.955 Tf 11.95 0 Td[(qrIy+r2Iyz+rpIxyrpIx)]TJ /F5 11.955 Tf 11.96 0 Td[(qrIxy)]TJ /F5 11.955 Tf 11.96 0 Td[(r2Ixz)]TJ /F5 11.955 Tf 11.95 0 Td[(rpIz+p2Ixz+qpIyzpqIy)]TJ /F5 11.955 Tf 11.95 0 Td[(rpIyz)]TJ /F5 11.955 Tf 11.96 0 Td[(p2Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(pqIx+q2Ixy+rqIxz377775(2{44) TheleftsideofEquation 2{44 canbeputintovectornotation,suchthatthevectorcomponentsoftheEarth-denedangularmomentumvector,HEB,aredenedbyindividualmomentterms,asshownbyEquation 2{45 .` dH dtEB=266664LMN377775EB(2{45) Afull-ordersetofnonlinearmomentequationscanbefoundbyrstinsertingEquation 2{45 intoEquation 2{44 andthenequatingsides,asseeninEquation 2{46 L=_pIx)]TJ /F5 11.955 Tf 11.96 0 Td[(qrIy+qrIz+(pr)]TJ /F1 11.955 Tf 14.11 0 Td[(_q)Ixy)]TJ /F1 11.955 Tf 11.95 0 Td[((pq+_r)Ixz+(r2)]TJ /F5 11.955 Tf 11.96 0 Td[(q2)Iyz+p_Ix)]TJ /F5 11.955 Tf 11.96 0 Td[(q_Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(r_IxzM=prIx+_qIy)]TJ /F5 11.955 Tf 11.96 0 Td[(prIz)]TJ /F1 11.955 Tf 11.95 0 Td[((qr)]TJ /F1 11.955 Tf 14.24 0 Td[(_p)Ixy+(p2)]TJ /F5 11.955 Tf 11.95 0 Td[(r2)Ixz+(pq)]TJ /F1 11.955 Tf 13.78 0 Td[(_r)Iyz+q_Iy)]TJ /F5 11.955 Tf 11.96 0 Td[(p_Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(r_IyzN=)]TJ /F5 11.955 Tf 9.3 0 Td[(pqIx+pqIy+_rIz+(q2)]TJ /F5 11.955 Tf 11.95 0 Td[(p2)Ixy+(qr)]TJ /F1 11.955 Tf 14.24 0 Td[(_p)Ixz)]TJ /F1 11.955 Tf 11.96 0 Td[((pr+_q)Iyz+r_Iz)]TJ /F5 11.955 Tf 11.95 0 Td[(p_Ixz)]TJ /F5 11.955 Tf 11.95 0 Td[(q_Iyz(2{46) 2.3.2KinematicEquations Thesixequationsofmotion,previouslydevelopedtorepresenttheforcesandmomentsactingontheaircraft,arenecessarybutnotsucient.Additionalequationsmustbeaddedinordertosolvetheoverallaircraftproblem.TheseadditionalequationsarenecessaryduetothefactthattheEuleranglesrepresentedintheforceequationscreatemorethansixunknowns. 2.3.2.1Orientationequations Threenewequationscanbeobtainedbyrelatingthethreebody-axissystemrates(p;q;r)tothethreeEulerrates(_ ;_;_).Itisnotedthatthisrelationshipcan 43
PAGE 44
beillustratedbyavectorequationwherethemagnitudeofthethreebodyratesequalsthethreeEulerratesand,asseeninEquation 2{47 !B=p^i+q^j+r^k=_ i+_j+_k(2{47) Toequatebothsidesofequation 2{24 ,itisnecessarythatbothvectorsarerepresentedinthesamecoordinateframe.Thus,bytransformingtheEuleranglesintothebody-xedcoordinatesystem,threenonlinearbodyrateequationscanbewritten,asseeninEquation 2{48 p=_)]TJ /F1 11.955 Tf 15.65 3.16 Td[(_ sinq=_cos+_ cossinr=_ coscos)]TJ /F1 11.955 Tf 14.2 3.15 Td[(_sin(2{48) ThesethreebodyrateequationcanalsobedenedintermsoftheEulerangles,asseeninEquation 2{49 _=p+q(sin+rcos)tan_=qcos)]TJ /F5 11.955 Tf 11.95 0 Td[(rsin_ =(qsin+rcos)sec(2{49) 2.3.2.2Positionequations Anadditionalthreeequationsarederivedfromtheightvelocitycomponentsrelativetotheinertiallydenedreferenceframe.Inordertoderivetheseequations,theinertiallydenedvelocitycomponents,representedinthebody-xedcoordinatesystem,mustrstbedened,asseeninEquation 2{50 266664xyz377775EB=266664dx/dtdy/dtdz/dt377775EB(2{50) 44
PAGE 45
AnEulertransformationmaybeusedtotransformthebody-denedvelocities,(u;v;w),intothedesiredEarth-denedvelocities.Thistransformationiscompletedbyapplyingequation 2{5 ,inreverseorder,tothebody-denedvelocities,asseeninEquation 2{51 266664dx/dtdy/dtdz/dt377775EB=266664coscos sinsincos )]TJ /F1 11.955 Tf 11.95 0 Td[(cossin cossincos +sinsin cossin sinsinsin )]TJ /F1 11.955 Tf 11.95 0 Td[(coscos cossinsin +sincos )]TJ /F1 11.955 Tf 11.29 0 Td[(sinsincoscoscos377775266664uvw377775B(2{51) Afull-ordersetofnon-linearvelocityequationscanthusbefoundbyrstcompletingthethematrixmultiplicationontherighthandsideofEquation 2{51 andthenequatingbothsides,asseeninEquation 2{52 _xEB=uBcoscos +vB(sinsincos )]TJ /F1 11.955 Tf 11.95 0 Td[(cossin )+wB(cossincos +sinsin )_yEB=uBcossin +vB(sinsinsin )]TJ /F1 11.955 Tf 11.95 0 Td[(coscos )+wB(cossinsin +sincos )_zEB=)]TJ /F5 11.955 Tf 9.3 0 Td[(uBsin+vB(sincos)+wBcoscos(2{52) IntegratingEquation 2{52 yieldstheairplane'spositionrelativetotheinertially-xedreferenceframe. 2.3.3TheEquationsCollected Thenonlinearaircraftequationsofmotioncanbecollectedintoaformalset,asshowninEquation 2{53 45
PAGE 46
m(_u+qw)]TJ /F5 11.955 Tf 11.95 0 Td[(rv)=)]TJ /F5 11.955 Tf 9.3 0 Td[(mgsin+()]TJ /F5 11.955 Tf 9.29 0 Td[(DcosA+LsinA)+TsinTm(_v+ru)]TJ /F5 11.955 Tf 11.95 0 Td[(pw)=mgsincos+FAy+FTym(_w+pv)]TJ /F5 11.955 Tf 11.96 0 Td[(qu)=FGz+FAz+FTzL=_pIx)]TJ /F5 11.955 Tf 11.96 0 Td[(qrIy+qrIz+(pr)]TJ /F1 11.955 Tf 14.11 0 Td[(_q)Ixy)]TJ /F1 11.955 Tf 11.95 0 Td[((pq+_r)Ixz+(r2)]TJ /F5 11.955 Tf 11.95 0 Td[(q2)Iyz+p_Ix)]TJ /F5 11.955 Tf 11.96 0 Td[(q_Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(r_IxzM=prIx+_qIy)]TJ /F5 11.955 Tf 11.96 0 Td[(prIz)]TJ /F1 11.955 Tf 11.95 0 Td[((qr)]TJ /F1 11.955 Tf 14.24 0 Td[(_p)Ixy+(p2)]TJ /F5 11.955 Tf 11.95 0 Td[(r2)Ixz+(pq)]TJ /F1 11.955 Tf 13.78 0 Td[(_r)Iyz+q_Iy)]TJ /F5 11.955 Tf 11.96 0 Td[(p_Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(r_IyzN=)]TJ /F5 11.955 Tf 9.3 0 Td[(pqIx+pqIy+_rIz+(q2)]TJ /F5 11.955 Tf 11.95 0 Td[(p2)Ixy+(qr)]TJ /F1 11.955 Tf 14.24 0 Td[(_p)Ixz)]TJ /F1 11.955 Tf 11.96 0 Td[((pr+_q)Iyz+r_Iz)]TJ /F5 11.955 Tf 11.95 0 Td[(p_Ixz)]TJ /F5 11.955 Tf 11.95 0 Td[(q_Iyz_=p+q(sin+rcos)tan_=qcos)]TJ /F5 11.955 Tf 11.96 0 Td[(rsin_ =(qsin+rcos)sec_xEB=uBcoscos +vB(sinsincos )]TJ /F1 11.955 Tf 11.95 0 Td[(cossin )+wB(cossincos +sinsin )_yEB=uBcossin +vB(sinsinsin )]TJ /F1 11.955 Tf 11.95 0 Td[(coscos )+wB(cossinsin +sincos )_zEB=)]TJ /F5 11.955 Tf 9.3 0 Td[(uBsin+vB(sincos)+wBcoscos(2{53) 2.4LinearizedEquationsofMotion Oftentimes,thenonlinearsetofmotionequationsislinearizedforuseinstabilityandcontrolanalysis.Thelinearizationiscarriedoutbymeansofthesmall-disturbancetheory.Whenusingthesmall-disturbancetheory,itisassumedthatthemotionoftheaircraftconsistsofsmalldeviationsfromareferenceconditionofsteadyight.Limitationsdoapplytothesmall-disturbancetheory,inthatproblemscontaininglargedisturbanceangles(i.e.:==2)cannotbelinearizedusingthismethod. Whenusingthesmall-disturbancetheory,allthevariablesintheequationsofmotionarereplacedbyareferencevalueplussomeperturbationasshowninEquation 2{54 46
PAGE 47
u=uo+up=p0+px=x0+xM=M0+Mv=v0+vq=q0+qy=y0+yN=N0+Nw=w0+wr=r0+rz=z0+zL=L0+L(2{54) Forconvenience,thereferenceightconditionisassumedtobesteadytrimmedightwithasymmetriccongurationandnoangularvelocity.TheseassumptionscanbephysicallyillustratedasshowninEquation 2{55 v0=p0=q0=r0=0=0=0(2{55) Furthermore,thex-axisisassumedtobeinitiallyalignedalongthedirectionoftheaircraft'svelocityvector,thus,w0=0.Asaresult,u0isequaltothereferenceightspeedand0tothereferenceangleofclimb.Forreasonsofsimplication,itisnotedthattheatrigonometricidentitymaybeapplied,asdenedinEquation 2{56 sin(0+)=sin0cos+cos0sin:=sin0+cos0cos(0+)=cos0cos)]TJ /F1 11.955 Tf 11.95 0 Td[(sin0sin:=cos0)]TJ /F1 11.955 Tf 11.96 0 Td[(sin0(2{56) Ageneralsetoflinearizedmotionequationsmaybeobtained,asshowninEquation 2{57 ,byapplyingthesmall-disturbancetheory,combinedwiththepreviouslymadeassumptions,tothenonlinearsetofmotionequations,givenbyequation 2{53 ,andretainingonlytherstorderterms. 47
PAGE 48
x0+x)]TJ /F5 11.955 Tf 11.95 0 Td[(mg(sin0+cos0)=m_uy0+y+mgcos0=m(_v+u0r)z0+z+mg(cos0)]TJ /F1 11.955 Tf 11.96 0 Td[(sin0)=m(_w)]TJ /F5 11.955 Tf 11.95 0 Td[(u0q)L0+L=Ix_p)]TJ /F5 11.955 Tf 11.95 0 Td[(Izx_rM0+M=Iy_qN0+N=)]TJ /F5 11.955 Tf 9.3 0 Td[(Izx_p+Iz_r_0+_=q_0_=p+rtan0_ 0_ =rsec0_xE0+_xE=(u0+u)cos0)]TJ /F5 11.955 Tf 11.95 0 Td[(u0sin0+wsin0_yE0+_yE=u0cos0+v_zE0+_zE=)]TJ /F1 11.955 Tf 9.3 0 Td[((u0+u)sin0)]TJ /F5 11.955 Tf 11.95 0 Td[(u0cos0+wcos0(2{57) IfallofthedisturbancesinEquation 2{57 aresetequaltozero,thentheresultingsetoflinearizedmotionequationsarerepresentativeofthosedenedforreferenceight. Iftheassumptionismadethattheaircraftisatitsreferenceightcondition,thedisturbancequantitiesareconsiderednegligibleandthereforesetequaltozero.Applyingthisassumptiontoequation 2{57 ,itisseenthatasetofequationsisdeveloped,asseeninEquation 2{58 ,whichcanbeusedtoeliminateallofthereferenceforcesandmomentsfoundinEquation 2{57 X0)]TJ /F5 11.955 Tf 11.95 0 Td[(mgsin0=0Y0=0Z0+mgcos0=0L0=M0+N0=0_xE0=u0cos0_yE0=0_zE0=)]TJ /F5 11.955 Tf 9.3 0 Td[(u0sin0(2{58) 48
PAGE 49
Equation 2{58 canbesustitutedbackintoEquation 2{57 ,suchthattheresultinglinearizedmotionequationscanberewrittenanddened,asseeninEquation 2{59 _u=x m)]TJ /F5 11.955 Tf 11.96 0 Td[(gcos0_v=y m+gcos0)]TJ /F5 11.955 Tf 11.95 0 Td[(u0r_w=z m)]TJ /F5 11.955 Tf 11.95 0 Td[(gsin0+u0qL=Ix_p)]TJ /F5 11.955 Tf 11.95 0 Td[(Izx_rM=Iy_qN=)]TJ /F5 11.955 Tf 9.3 0 Td[(Izx_q_=q_=p+rtan0_=rsec0_xE=ucos0)]TJ /F5 11.955 Tf 11.96 0 Td[(u0sin0+wsin0_yE=u0cos0+v_zE=)]TJ /F1 11.955 Tf 9.3 0 Td[(usin0)]TJ /F5 11.955 Tf 11.96 0 Td[(u0cos0+wcos0(2{59) TheperturbationtermsrepresentaerodynamicforcesandmomentsthatcanbeexpressedbymeansofaTaylorseriesexpansion.TheTaylorseriesexpansionmaycontainallofthemotionvariables,butisnormallyreducedtoonlythesignicanttermsrelevanttothatpaticularforceormoment.Forexample,theTaylorseriesexpansionforthechangeinrollmoment,L,maybeexpressedasafunctionofthemoments,forcesandcontrolsurfacedeections,asseeninEquation 2{60 L=@L @uu+@L @vv+@L @ww+@L @qq+@L @pp+@L @rr+@L @aa+@L @rr+@L @ee(2{60) 2.5Examples Toillustratethederivationandlinearizationprocessfurther,anexamplewillbegiven.Theexamplewillexamineanaircraftthatiscapableofmorphingboth 49
PAGE 50
symmetricallyandasymmetrically.Itisnoticedthattheonlyequationsthatvarywithsymmetryarethemomentequationsgivenbyequation 2{46 .Thenonlinearmomentequationswillrstbelinearizedandthenreducedaccordingtocongurationandaerodynamicassumptions. 2.5.1Linearization Startingwiththenonlinearmomentequations,asdenedinequation 2{46 ,thesmall-disturbancetheoryisapplied,thusresultingintheperturbationequationsshowninEquation 2{61 L0+L=_pIx+(r0q+q0r)(Iz)]TJ /F5 11.955 Tf 11.95 0 Td[(Iy)+(p0r+r0p)]TJ /F1 11.955 Tf 11.96 0 Td[(_q)Ixy)]TJ /F1 11.955 Tf 11.96 0 Td[((p0q+q0p+_r)Ixz+(2r0)]TJ /F1 11.955 Tf 11.96 0 Td[(2q0q)Iyz+p_Ix)]TJ /F1 11.955 Tf 11.96 0 Td[(q_Ixy)]TJ /F1 11.955 Tf 11.95 0 Td[(r_IxzM0+M=(r0p+p0r)(Ix)]TJ /F5 11.955 Tf 11.96 0 Td[(Iz)+_qIy)]TJ /F1 11.955 Tf 11.95 0 Td[((q0r+r0q+_p)Ixy+(2p0p)]TJ /F1 11.955 Tf 11.96 0 Td[(2r0r)Ixz+(q0p+p0q)]TJ /F1 11.955 Tf 11.95 0 Td[(_r)Iyz+q_Iy)]TJ /F1 11.955 Tf 11.95 0 Td[(p_Ixy)]TJ /F1 11.955 Tf 11.96 0 Td[(r_IyzN0+N=(p0q+q0p)(Iy)]TJ /F5 11.955 Tf 11.95 0 Td[(Ix)+_rIz)]TJ /F1 11.955 Tf 11.96 0 Td[((2q0q)]TJ /F1 11.955 Tf 11.96 0 Td[(2p0p)Ixy+(r0q)]TJ /F5 11.955 Tf 11.96 0 Td[(q0r)]TJ /F1 11.955 Tf 11.96 0 Td[(_p)Ixz)]TJ /F1 11.955 Tf 11.96 0 Td[((r0p+p0r)]TJ /F1 11.955 Tf 11.95 0 Td[(_q)Iyz+r_Iz)]TJ /F1 11.955 Tf 11.96 0 Td[(p_Ixz)]TJ /F1 11.955 Tf 11.95 0 Td[(q_Iyz(2{61) Equation 2{61 canberearrangedinsuchamannerthatitisexpressedasadierentialequation,asseeninEquation 2{62 Ix_p)]TJ /F5 11.955 Tf 11.96 0 Td[(Ixy_q)]TJ /F5 11.955 Tf 11.95 0 Td[(Ixz_r=L0+L+()]TJ /F5 11.955 Tf 9.3 0 Td[(r0Ixy+q0Ixz)]TJ /F1 11.955 Tf 14.68 3.02 Td[(_Ix)p+()]TJ /F5 11.955 Tf 9.3 0 Td[(r0(Iz)]TJ /F5 11.955 Tf 11.95 0 Td[(Iy)+p0Ixz+2q0Iyz+_Ixy)q+()]TJ /F5 11.955 Tf 9.3 0 Td[(q0(Iz)]TJ /F5 11.955 Tf 11.96 0 Td[(Iy))]TJ /F5 11.955 Tf 11.96 0 Td[(p0Ixy)]TJ /F1 11.955 Tf 11.96 0 Td[(2r0Iyz+_Ixz)r)]TJ /F5 11.955 Tf 9.29 0 Td[(Ixy_p)]TJ /F5 11.955 Tf 11.96 0 Td[(Iy_q)]TJ /F5 11.955 Tf 11.95 0 Td[(Iyz_r=M0+M+()]TJ /F5 11.955 Tf 9.3 0 Td[(r0(Ix)]TJ /F5 11.955 Tf 11.96 0 Td[(Iz))]TJ /F1 11.955 Tf 11.96 0 Td[(2p0Ixz)]TJ /F5 11.955 Tf 11.95 0 Td[(q0Iyz+_Ixy)p+(r0Ixy)]TJ /F5 11.955 Tf 11.95 0 Td[(p0Iyz)]TJ /F1 11.955 Tf 14.68 3.02 Td[(_Iy)q+()]TJ /F5 11.955 Tf 9.3 0 Td[(p0(Ix)]TJ /F5 11.955 Tf 11.96 0 Td[(Iz)+q0Ixy+2r0Ixz+_Iyz)r)]TJ /F5 11.955 Tf 9.29 0 Td[(Ixz_p)]TJ /F5 11.955 Tf 11.96 0 Td[(Iyz_q+Iz_r=N0+N+()]TJ /F5 11.955 Tf 9.3 0 Td[(q0(Iy)]TJ /F5 11.955 Tf 11.96 0 Td[(Ix)+2p0Ixy+r0Iyz+_Ixz)p+()]TJ /F5 11.955 Tf 9.3 0 Td[(p0(Iy)]TJ /F5 11.955 Tf 11.95 0 Td[(Ix))]TJ /F5 11.955 Tf 11.96 0 Td[(r0Ixz)]TJ /F1 11.955 Tf 11.96 0 Td[(2q0Ixy+_Iyz)q+()]TJ /F5 11.955 Tf 9.3 0 Td[(q0Ixz+p0Iyz)]TJ /F1 11.955 Tf 14.68 3.02 Td[(_Iz)r(2{62) 50
PAGE 51
Theaircraftisassumedtobeatstraightandleveltrimmed(reference)ight,therefore,thedisturbancevaluescanbesetequaltozeroandEquation 2{62 canbefurtherreduced,asshownbyEquation 2{63 Ix_p)]TJ /F5 11.955 Tf 11.95 0 Td[(Ixy_q)]TJ /F5 11.955 Tf 11.96 0 Td[(Ixz_r=L)]TJ /F1 11.955 Tf 14.68 3.02 Td[(_Ixp+_Ixyq+_Ixzr)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixy_p+Iy_q)]TJ /F5 11.955 Tf 11.96 0 Td[(Iyz_r=M+_Ixyp)]TJ /F1 11.955 Tf 14.68 3.02 Td[(_Iyq+_Iyzr)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixz_p)]TJ /F5 11.955 Tf 11.96 0 Td[(Iyz_q+Iz_r=N+_Ixzp+_Iyzq)]TJ /F1 11.955 Tf 14.69 3.03 Td[(_Izr(2{63) Notethattheinertialrates(_Iterms)areretainedinthelinearizedmomentequations.Theretentionofthesetermsisdoneinordertoaccountforthefactthattheaircraftiscapableofmorphing.Solvingequation 2{63 intermsof_p;_qand_rresultsinthethreeequationsshowninEquation 2{64 _p=Ppp+Pqq+Prr+PLL+PMM+PNN D_q=Qpp+Qqq+Qrr+QLL+QMM+QNN D_r=Rpp+Rqq+Rrr+RLL+RMM+RNN D(2{64) ThecoecientsoftheperturbationtermsinEquation 2{63 areexpressedasfunctionsoftheinertialmoments,productsandrates,asseeninEquation 2{65 51
PAGE 52
Pp=)]TJ /F5 11.955 Tf 9.3 0 Td[(I2yz_Ix+IzIy_Ix)]TJ /F5 11.955 Tf 11.96 0 Td[(IxyIz_Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(IxyIyz_Ixz)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIyz_Ixy)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIy_IxzPq=IxyIz_Iy)]TJ /F5 11.955 Tf 11.95 0 Td[(IxyIyz_Iyz+IxzIyz_Iy)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIy_Iyz)]TJ /F5 11.955 Tf 11.95 0 Td[(IzIy_Ixy+Iyz_IxyPr=)]TJ /F5 11.955 Tf 9.3 0 Td[(IxyIz_Iyz+IxyIyz_Iz+I2yz_Ixz+IxzIy_Iz)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIyz_Iyz)]TJ /F5 11.955 Tf 11.95 0 Td[(IyIz_IxzPL=I2yz)]TJ /F5 11.955 Tf 11.95 0 Td[(IzIyPM=)]TJ /F5 11.955 Tf 9.29 0 Td[(IxyIz)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIyzPN=)]TJ /F5 11.955 Tf 9.29 0 Td[(IxyIyz)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIyQp=)]TJ /F5 11.955 Tf 9.3 0 Td[(IxIz_Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(IxIyz_Ixz)]TJ /F5 11.955 Tf 11.96 0 Td[(IxzIxy_Ixz+IyzIxz_Ix+I2xz_Ixy+IxyIz_IxQq=IxIz_Iy)]TJ /F5 11.955 Tf 11.96 0 Td[(IxIyz_Iyz)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIxy_Iyz)]TJ /F5 11.955 Tf 11.96 0 Td[(IyzIxz_Ixy)]TJ /F5 11.955 Tf 11.95 0 Td[(I2xz_Iy)]TJ /F5 11.955 Tf 11.96 0 Td[(IxyIz_IxyQr=)]TJ /F5 11.955 Tf 9.3 0 Td[(IxIz_Iyz+IxIyz_Iz+IxzIxy_Iz)]TJ /F5 11.955 Tf 11.96 0 Td[(IyzIxz_Ixz+I2xz_Iyz)]TJ /F5 11.955 Tf 11.96 0 Td[(IxyIz_IxzQL=)]TJ /F5 11.955 Tf 9.3 0 Td[(IyzIxz)]TJ /F5 11.955 Tf 11.95 0 Td[(IxyIzQM=)]TJ /F5 11.955 Tf 9.3 0 Td[(IxIz+I2xzQN=)]TJ /F5 11.955 Tf 9.3 0 Td[(IxIyz)]TJ /F5 11.955 Tf 11.96 0 Td[(IxzIxyRp=I2xy_Ixz+IxzIy_Ix)]TJ /F5 11.955 Tf 11.96 0 Td[(IxIyz_Ixy+IyzIxy_Ix)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIxy_Ixy)]TJ /F5 11.955 Tf 11.96 0 Td[(IxIy_IxzRq=)]TJ /F5 11.955 Tf 9.29 0 Td[(IxzIy_Ixy+I2xy_Iyz+IxIyz_Iy)]TJ /F5 11.955 Tf 11.96 0 Td[(IyzIxy_Ixy+IxzIxy_Iy)]TJ /F5 11.955 Tf 11.96 0 Td[(IxIy_IyzRr=)]TJ /F5 11.955 Tf 9.29 0 Td[(IxzIy_Ixz)]TJ /F5 11.955 Tf 11.96 0 Td[(I2xy_Iz)]TJ /F5 11.955 Tf 11.95 0 Td[(IyzIxy_Ixz+IxIy_Iz)]TJ /F5 11.955 Tf 11.95 0 Td[(IxzIxy_Iyz)]TJ /F5 11.955 Tf 11.96 0 Td[(IxIyz_IyzRL=)]TJ /F5 11.955 Tf 9.3 0 Td[(IxzIy)]TJ /F5 11.955 Tf 11.95 0 Td[(IyzIxyRM=)]TJ /F5 11.955 Tf 9.3 0 Td[(IxzIxy)]TJ /F5 11.955 Tf 11.95 0 Td[(IxIyzRN=)]TJ /F5 11.955 Tf 9.3 0 Td[(IxIy+I2xy(2{65) Thecommondenominator,D,foundinEquation 2{64 isalsoexpressedasafunctionoftheinertialmoments,products,andrates,asseeninEquation 2{66 D=)]TJ /F5 11.955 Tf 9.29 0 Td[(IxIyIz+I2xzIy+IzI2xy+IxI2yz+2IyzIxyIxz(2{66) MotionvariablescanbeaccountedforinthemomentequationsbyexpressingthemasatermintheTaylorseriesexpansion,asseeninEquation 2{67 52
PAGE 53
L=@L @uu+@L @vv+@L @ww+@L @qq+@L @pp+@L @rr+@L @aa+@L @rr+@L @eeM=@M @uu+@M @vv+@M @ww+@M @qq+@M @pp+@M @rr+@M @aa+@M @rr+@M @eeN=@N @uu+@N @vv+@N @ww+@N @qq+@N @pp+@N @rr+@N @aa+@N @rr+@N @ee(2{67) ThefullyexpandedsetoflinearizedmomentequationscanbeobtainedbyinsertingEquations 2{63 and 2{67 intoequation 2{63 andmultiplyingouttheterms. 2.5.2AsymmetricMorphing Figure2-7. AsymmetricCongurations Theasymmetriccaseassumesthatthereisnosymmetrytakenwithrespecttotheaircraft'scenterofgravity.Itisalsoassumedthattheaircraftisactivelymorphing,andtherefore,theinertialratesareretained.Themomentequationsspecictotheasymmetricmorphingcase,havepreviouslybeendenedandareshowninEquations 2{63 2{67 .Ifitisdeterminedthattheaircraftnolongermorphs,theinertialratesgotozeroandequation 2{63 canbefurtherreduced,asseeninEquation 2{68 Ix_p)]TJ /F5 11.955 Tf 11.96 0 Td[(Ixy_q)]TJ /F5 11.955 Tf 11.95 0 Td[(Ixz_r=L)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixy_p+Iy_q)]TJ /F5 11.955 Tf 11.95 0 Td[(Iyz_r=M)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixz_p)]TJ /F5 11.955 Tf 11.95 0 Td[(Iyz_q+Iz_r=N(2{68) 53
PAGE 54
Figure2-8. SymmetricCongurations 2.5.3SymmetricConguration Thesymmetriccaseassumesthatthereissymmetryinthexy-andyz-planes.Asadirectresult,allofthetermsrelateddescribedbyIxyandIyzgotozero.Thesymmetricexampleissimilartoasymmetriccase,inthatitisalsoassumedtobeactivelymorphing.Therefore,theinertialratesareagainretainedandtheresultingmomentequationsforsymmetricmorphingcanbeshownbyEquation 2{69 Ix_p)]TJ /F5 11.955 Tf 11.96 0 Td[(Ixz_r=L)]TJ /F1 11.955 Tf 14.68 3.02 Td[(_Ixp+_Ixzr)]TJ /F5 11.955 Tf 9.3 0 Td[(Iy_q=M)]TJ /F1 11.955 Tf 14.68 3.03 Td[(_Iyq)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixz_p+Iz_r=N+_Ixzp)]TJ /F1 11.955 Tf 14.69 3.02 Td[(_Izr(2{69) Ifitisdeterminedthattheaircraftnolongermorphs,theinertialratesagaingotozeroandequation 2{69 canbereducedtothemomentequations,asseeninEquations 2{59 and 2{70 Ix_p)]TJ /F5 11.955 Tf 11.96 0 Td[(Ixz_r=LIy_q=M)]TJ /F5 11.955 Tf 9.3 0 Td[(Ixz_p+Iz_r=N(2{70) 54
PAGE 55
2.6ATechnicalApproachtotheEquationsofMotion Theequationsofmotion,includingasymmetricandtime-varyingshapesofassociatedmorphingactuation,canbesimpliedintotheformgiveninEquation 2{71 .Thisformdescribesasystemwithnxstatesalongwithnmorphingactuatorsandnutraditionalactuatorssuchthatx2Rnxisthestatevector,2Rnisthecommandtomorphing,andu2Rnuisthecommandtotraditionalcontrolsurfaces.ThecoecientsintheequationsofmotionarecombinedintoA2RnxnxandB2Rnxnuwhicharefunctionsofboththemorphingconguration,,andtheightcondition,. _x=A(;)x+B(;)u(2{71) ThemodelinEquation 2{72a representstheightdynamicsforanaircraftwithaxedconguration.Inotherwords,thismodeldescribesanaircraftwithoutmorphing.Thismodelutilizesmatricesthatdependonanexogenous,butmeasurable,variablesuchasMachoraltitude.Thisformisactuallythelinearparameter-varying(LPV)dynamicswhichhavebeenextensivelystudied[ 42 ]. _x=A()x+B()uu=K()x(2{72a) ThemodelinEquation 2{72b representstheightdynamicsforanaircraftataxedightcondition.Theaircraftisallowedtomorph;however,theaerodynamicsdonotvarybecauseofMachoraltitudechanges.Theresearcherproposestodenotethisformaslinearinput-varyingdynamics(LIV)becauseofthedependencyonthecommandinputfordynamicswhicharelinearinthestates. _x=A()x+B()uu=K()x(2{72b) 55
PAGE 56
Clearlyamorphingaircrafthashighlystructureddynamics.Inparticular,notethattheightdynamicsreverttoalinearstate-spaceforminEquation 2{72c whenbothmorphingandightconditionarexed.Themodelthusrepresentsanimportantphysicalaspectofmorphing;specically,theaircraftwillhavetraditionaldynamicsformaneuveringaroundtrimwhenmorphedintoaxedconguration. _x=Ax+Buu=Kx(2{72c) 56
PAGE 57
CHAPTER3LINEARTIME-VARYINGEIGENSTRUCTURESANDTHEIRSTABILITY:ASURVEY 3.1DenitionofanLTVSystem Alinear-time-varying(LTV)systemcanbedenedbythecoupledlinearhomogeneousdierentialequationsshowninEquation 3{1 _x=A(t)x(t)+B(t)u(t)y(t)=C(t)x(t)+D(t)u(t)(3{1) Forthepurposesofstudyinthispaper,theLTVstateequationwillberestrictedtoformshowninEquation 3{2 _x=A(t)x(t)(3{2) FromEquation 3{2 ,itisnotedthatcoecientmatrix,A(t),iscomposedsuchthatA2R+!Rnxn,whilethestatevector,x(t),iscomposedsuchthatx2R+!Rn,andnisthenumberofstatesinthesystem.Usingthisdenitionofalineartime-varyingsystem,multipleconceptsforLTVeigenpairsandtheirstabilityhavebeendeveloped. 3.2Kamen'sConceptofPolesofanLTVSystem 3.2.1TwoStateSystem ConsiderasecondorderLTVsystemoftheform: x+a1(t)_x+a0(t)x(t)=0(3{3) whichcanbewrittenusingoperatornotationD=d dt,as (D2+a1(t)D+a0(t))x(t)=0(3{4) Ifthereexistfunctionsp1(t)andp2(t)suchthatonecanwrite (D2+a1(t)D+a0(t))x(t)=(D)]TJ /F5 11.955 Tf 11.95 0 Td[(p1(t))f(D)]TJ /F5 11.955 Tf 11.95 0 Td[(p2(t))x(t)g(3{5) 57
PAGE 58
andwedenea(noncommutative)polynomialmultiplicationosuchthat f(D)]TJ /F5 11.955 Tf 11.96 0 Td[(p1(t))o(D)]TJ /F5 11.955 Tf 11.95 0 Td[(p2(t))gx(t)=(D)]TJ /F5 11.955 Tf 11.96 0 Td[(p1(t))f(D)]TJ /F5 11.955 Tf 11.96 0 Td[(p2(t))x(t)g(3{6) thencomparing 3{5 and 3{6 ,onecandeneDop2(t)=p2(t)D+_p2(t),andnallyarriveatanequationforp2as p22(t)+_p2(t)+a1(t)p2(t)+a0(t)=0(3{7) fromwhichonecanthendeterminep1using p1(t)+p2(t)=)]TJ /F5 11.955 Tf 9.3 0 Td[(a1(t)(3{8)(p1(t);p2(t))thenformapoleset,andp2(t)iscalledarightpole.Notethatthisisanorderedpoleset,andevenwhenp1(t)andp2(t)arecomplex,theyneednot,ingeneral,becomplexconjugate[ 59 ]. Themodeassociatedwitharightpolep2(t)isdenedas p2(t;0)=eRt0p2()d(3{9) andonecanwrite x(t)=C121(t;0)+C222(t;0)(3{10) wherep21andp22representtworightpolesdeterminedbysolvingequation 3{7 fromtwodierentinitialconditions(whicharetypicallythetwotimeinvariantpolesatt=0,alsoreferredtoasthe"frozen-time"rootsof 3{3 att=0);andC1andC2areconstants.21(t;0)and22(t;0)giveinformationaboutthestabilityoftheLTVsystem[ 59 ]. 3.2.2FourStateSystem Thegeneralizedformofa4-statesystemisexpressedinEquation 3{11 .Inthiscase,thederivativesofthestate,w(t),aremultipliedbyrealcoecientsofA0(t);A1(t);A2(t);A3(t).Theexpressionisalteredusingoperatornotation,givenasD=d dtasinEquation 3{4 ,togenerateEquation 3{12 58
PAGE 59
0=d4w dt4+A3(t)d3w dt3+A2(t)d2w dt2+A1(t)dw dt+A0(t)w (3{11) =)]TJ /F5 11.955 Tf 5.48 -9.68 Td[(D4+A3(t)D3+A2(t)D2+A1(t)D+A0(t)w(t) (3{12) Theconceptofanordersetofpoles,(p1(t);p2(t);p3(t)p4(t)),isintroducedandrelatetothedynamicsasinEquation 3{13 )]TJ /F5 11.955 Tf 5.48 -9.69 Td[(D4+A3(t)D3+A2(t)D2+A1(t)D+A0(t)w(t)=(D)]TJ /F5 11.955 Tf 9.3 0 Td[(p1(t))[(D)]TJ /F5 11.955 Tf 9.3 0 Td[(p2(t)f(D)]TJ /F5 11.955 Tf 9.3 0 Td[(p3(t))[D)]TJ /F5 11.955 Tf 9.3 0 Td[(p4(t)]g]w(3{13) Therightpole,whichistheonlypoleneededtofullycharacterizethesystemisgeneratedasasolutiontoEquation 3{14 .Notethatagaintherightpole,p4,actuallyhas4valuesofp4iresultingfromchoiceofinitialconditionsforthepoles. 0=d3p4i dt3+(4p4i(t)+A3(t))d2p4i dt2+(6p24i(t)+A2(t)+3A3(t)p4i(t))dp4i dt+3(dp4i dt)2+p44i(t)+A3(t)p34i(t)+A2(t)p24i(t)+A1(t)p4i(t)+A0(t) (3{14) Asetofeigenvectorsareagainassociatedwitheachpole.Eacheigenvector,vi,anditassociatedpole,p4i,mustsatisfyEquation 3{15 (A(t))]TJ /F5 11.955 Tf 11.95 0 Td[(p4i(t))vi(t)=_vi(t)(3{15) Thestatesofthesystemarecomputedasalinearcombinationoftheseeigenvectorsandthemodes,givenasi=exp(Rt0p4i(t)),associatedwitheachpole.Theresultingexpressionisgivenalongwiththescalarconstants,Ci,inEquation 3{16 x(t)=C1v1(t)41(t)+C2v2(t)42(t)+C3v3(t)43(t)+C4v4(t)44(t)(3{16) 59
PAGE 60
ThedecompositionofthestatesintotheformofEquation 3{16 indicatesthetime-varyingparametersessentiallydiagonalizethesystem.ConsiderthatamatrixdenedasV(t)=[v1(t)jv2(t)jv3(t)jv4(t)]willdiagonalizethesystemmatrixaslongasV(t)isinvertibleandbounded. ThestabilityofthesystemisdeterminedbytherelationshipinEquation 3{16 .Essentially,thesystemhasasymptoticstabilityforwhichstateswilltendtoequilibriumifandonlyifthemagnitudeofthemodegoestozeroastimegoesincreases.Thiscondition,whichisexpressedasj4ij!0ast!1foreachi=1;2;3;4,isequivalenttoaconditionontherealpartofthepolebeingR10pR()d<0. 3.3ZhuandJohnson Annth-order,scalar,lineartime-varying(LTV)system y(n)=Xnk=1k(t)y(k)]TJ /F7 7.97 Tf 6.59 0 Td[(1)(3{17) canbeconvenientlyrepresentedasDfyg=0usingthescalarpolynomialdierentialoperator(SPDO) D=n+Xnk=1k(t)k)]TJ /F7 7.97 Tf 6.59 0 Td[(1(3{18) where=d=dtisthederivativeoperatordenedonthedierentialring(D-ring),K. UsingatechniquedevelopedbyFloquet[ 39 ],theSPDOcanbefactoredinto D=()]TJ /F5 11.955 Tf 11.96 0 Td[(n(t))()]TJ /F5 11.955 Tf 11.96 0 Td[(2(t))()]TJ /F5 11.955 Tf 11.96 0 Td[(1(t))(3{19) IfDisdenedasanSPDOoperatorwithtime-varyingcoecients,k2K,k=1;2;:::;n,thenthescalarfunctionsk2K,k=1;2;:::;ngivenbythefactorizationofequation 3{19 arecalledSeriesD-eigenvalues(SD-eigenvalues)ofD.Moreover,ifthereexistssome(t)suchthat(t)=1(t),then(t)iscalledaParallelD-eigenvalue(PD-eigenvalues)ofD. 60
PAGE 61
Amulti-setofSD-eigenvalues,)]TJ /F11 7.97 Tf 165.63 -1.79 Td[(=fk(t)gnk=1,canbedenedasaSeriesD-spectrum(SD-spectrum)forDifk(t)satisesequation 3{19 ;whereasamulti-setofPD-eigenvalues=fk(t)gnk=1canbedenedasaParallelD-spectrum(PD-spectrum)forDifk(t)arePD-eigenvaluesforDandfyk=exp(Rk(t)dt)gnk=1constitutesafundamentalsetofsolutionstoDfyg=0. Assumingthatfk(t)gnk=1isaSD-spectrumforannth-orderSPDOD,thenthePD-spectrumfk(t)gnk=1forDcanbedirectlyobtainedby k(t)=1(t)+_qk(t)q)]TJ /F7 7.97 Tf 6.58 0 Td[(1k(t);k=2;3;:::;n(3{20) where, qk(t)=Z21(t)Z32(t)ZZk;k)]TJ /F7 7.97 Tf 6.59 0 Td[(1(t)dk)]TJ /F7 7.97 Tf 6.59 0 Td[(1t(3{21) and ij(t)=eR(i(t))]TJ /F11 7.97 Tf 6.58 0 Td[(j(t))dt(3{22) Thefactthatascalarequationcanalwaysbetransformedintoanequivalentcompanioncanonicalvectorequation,_x=Ac(t)x,leadsonetodenethecompanionmatrixassociatedwithD,Ac(t)as Ac(t)=2666666666640100...........................0001)]TJ /F5 11.955 Tf 9.3 0 Td[(1)]TJ /F5 11.955 Tf 9.3 0 Td[(2)]TJ /F5 11.955 Tf 61.1 0 Td[(n377777777775(3{23) whichdirectlyresultsinthematrices 61
PAGE 62
\(t)=2666666666641(t)10002(t).....................0............100n(t)377777777775(3{24) and (t)=diag1(t)2(t)n(t)(3{25) denedasaSeriesSpectralcanonicalform(SScanonicalform)andaParallelSpectralcanonicalform(PS-canonicalform),repectively.ItisimportanttonotethatbothmatricesaredenedoverthebasisdescribedbyDandAc(t),andthatforanytime-invariantD,therealwaysexistsanSD-spectrumoftime-invariantisatisfyingequation 3{19 Welldened(freeofnite-timesingularities)SD-andPD-spectracanbeusedtoderiveanalyticalsolutionsandstabilitycriteriaforLTVsystemsinawaysimilartothoseforLTIsystems.Inparticular,iffk(t)gnk=1isawell-denedPD-spectrumforD,thenthegeneralsolutiony(t)toDfyg=0isgivenby y(t)=nXi=1CieRi(t)dt whereCiareconstantsofintegrationdeterminedbyinitialconditions. 3.3.1GeneralizedPD-Eigenvectors AssumingthatDisannth-orderSPDOwithaPD-spectrumfk(t)gnk=1andthatfyi(t)gnk=1isafundamentalsetofsolutionsforDfyg=0,suchthatyi(t)=eR(i(t)dt),thenthematrix,D,canbedenotedbythediagonalmatrix D=diag[y1y2:::yn](3{26) 62
PAGE 63
Then WD)]TJ /F7 7.97 Tf 6.58 0 Td[(1=V(12:::n)=266666666664111D1f1gD2f1gDnf1gD21f1gD22f1gD2nf1g............Dn)]TJ /F7 7.97 Tf 6.58 0 Td[(11f1gDn)]TJ /F7 7.97 Tf 6.59 0 Td[(1nf1g377777777775(3{27) and detV=detWnk=1y)]TJ /F7 7.97 Tf 6.59 0 Td[(1k(3{28) whereD1=(+i),Dki=DiDk)]TJ /F7 7.97 Tf 6.58 0 Td[(1i,andW=W(y1y2:::yn)istheWronskianmatrixassociatedwithfyigni=1. ThecanonicalcoordinatetransformationmatrixV(t)iscalledthemodalcanonicalmatrixforDassociatedwiththePD-spectrumfigni=1andthedeterminantinequation 3{28 iscalledtheassociatedmodaldeterminant.Itisnotedthatthecolumnvectorsvi(t)ofV(t)satisfy Ac(t)vi(t))]TJ /F5 11.955 Tf 11.96 0 Td[(i(t)vi(t)=_vi(t)(3{29) andtherowvectorsuTi(t)ofV(t)satisfy UTi(t)Ac(t))]TJ /F5 11.955 Tf 11.96 0 Td[(i(t)UTi(t)=_UTi(t)(3{30) Thus,vi(t)anduTi(t)havebeencalledcolumnPD-eigenvectorsandrowPD-eigenvectors,respectively,ofDassociatedwithit. 3.3.2StabilityCriteria AssumingthatDisawell-denednth-orderSPDOwithawell-denedPD-spectrumfk(t)gnk=1inI=[T0;1),suchthatvk(t)anduTi(t)arecolumnPD-eigenvectorsandrow 63
PAGE 64
PD-eigenvectorsassociatedwithk(t),respectively.ThenthenullsolutiontoDfyg=0isuniformlyasymptoticallystableforallt0T0ifandonlyif thereexistsa0 0and0
PAGE 65
issaidtobethemode-vectorofA(t)associatedwiththex-eigenpairfi(t);ui(t)gandisusedtodeterminethestabilityofthelineartime-varyingsystem. IfA(t)hasndistincteigenvalues,thenA(t)canalwaysbediagonalizedbyanon-singularmatrixformedbytheeigenvectorsofA(t).Thus,thex-eigenvectorsofA(t)arefoundbyusingtheeigenvectorsofA(t)andthederivedmatricesobtainedfromA(t). Toobtainthex-eigenpairsofA(t),itisrstassumedthatj=1andAj(t)=A(t).TheeigenpairsofA(t),fji(t);uji(t)gi=1;2;:::;narethencomputedtocheckfordistinctnessoftheeigenvalues,fji(t)gni. Iftheeigenvaluesaredistinctthen Sj(t)=[u1j(t)u2j(t):::unj(t)](3{34) where j(t)=diag[1j(t)2j(t):::nj(t)](3{35) butiftheeigenvalues,fji(t)gni,arenotdistinct,thenanon-singularmatrixSj(t)ischosen.OnceSj(t)andj(t)hasbeendetermined,thedierentialequation Ej(t)=_Sj(t)S)]TJ /F7 7.97 Tf 6.59 0 Td[(1j(t)+Aj(t))]TJ /F9 11.955 Tf 11.95 0 Td[(A(t)(3{36) issolved.ForcaseswhereEj(t)0,thenSj(t)=S(t)andj(t)=(t).Thecolumns,ui(t)ofS(t)givethex-eigenvectorsofA(t)andtheelementsof(t),fi(t);ui(t)g,givethex-eigenvaluesofA(t).ForcaseswhereEj(t)6=0,thenj=j+1andAj(t)=Aj)]TJ /F7 7.97 Tf 6.58 0 Td[(1(t))]TJ /F9 11.955 Tf 11.45 0 Td[(Ej)]TJ /F7 7.97 Tf 6.59 0 Td[(1(t).TheprocedureisthencontinuedbyrecalculatingtheeigenpairsofA(t),fji(t);uji(t)gi=1;2;:::;nandcheckingfortheirdistinctness.TheiterationprocessesendswhenEj(t)0. Thecalulatedx-eigenpaircanthenbeusedtoproduceasystemresponsesuchthat 65
PAGE 66
x(t)=C1eR1(t)dte1(t)+C2eR2(t)dte2(t) whereC1andC2arecontantsofintegrationdeterminedbyinitialconditions. 3.4.2StabilityCriteria Anecessaryconditionforasymptoticstabilityandthesucientconditionfortheinstabilityofthesystem,equation 3{31 ,canbeexplicitlyanddirectlycheckedformthesystemmatrixA(t). Thenecessaryconditionforthesystem,equation 3{31 ,tobeasymptoticallystable,isthatforanyt>t0,tRt0trA()d!t!1.WheretristhetraceofA(t)andisrepresentedasA(t)=nPi=1aii(t). Thesucientconditionforinstabilityofthesystem,equation 3{31 ,isthatforanyt>t0,tRt0trA()d!1t!1. Necessaryandsucientconditionsforasymptoticstabilityandinstability,respectively,canalsobedescribedforthelineartime-varyingsystem y(n)+a1(t)y(n)]TJ /F7 7.97 Tf 6.59 0 Td[(1)+:::+an(t)y=0(3{37) Asucientconditionforinstabilityofthesystem,equation 3{37 ,isdescribedby tZtoa1()d!t!1(3{38) whereanecessaryconditionforasymptoticstabilityisdescribedby tZtoa1()d!1t!1(3{39) Thenecessaryandsucientconditionspresentedthusfaronlyprovidepartialanswerstothestabilityoflineartime-varyingsystems.Therefore,necessaryandsucientconditionsforthestabilityoflineartime-varyingsystems,intermsofmode-vectors, 66
PAGE 67
aredeveloped.Recallthatthemode-vectorsarenon-zerodierentiablevectorsofA(t),associatedwiththex-eigenpairfi(t);ui(t)g. Thelineartime-varyingsystem, 3{31 ,isstableifandonlyifeverymode-vectormi(t)ofA(t)isboundedsuchthat kmi(t)k<1t>t0;i=1;2;:::;n(3{40) andasymptoticallystableifandonlyif,inadditiontoequation 3{39 kmi(t)k!0t!1;i=1;2;:::;n(3{41) 3.5O'BrienandIglesias 3.5.1Formulation Consideralineartime-varyingsystemasdescribedinEquation 3{42 forsomestatevector,x(t)2Rn,andassociatedmatrix,A(t)2Rnn,forallt2R. _x(t)=A(t)x(t)(3{42) Introduceanewvectorofstates,z(t)2Rn,whichrelatestox(t)throughatransformation,S(t)2Rnn,forallt2R.Therelationshipresultsasz(t)=S)]TJ /F7 7.97 Tf 6.59 0 Td[(1(t)x(t).TheexpressioninEquation 3{42 forx(t)isthuswrittenintermsofz(t)inEquation 3{43 _S(t)z(t)+S(t)_z(t)=A(t)S(t)z(t) (3{43) S(t)_z(t)=A(t)S(t)z(t))]TJ /F1 11.955 Tf 15.25 3.02 Td[(_S(t)z(t) (3{44) _z(t)=S)]TJ /F7 7.97 Tf 6.59 0 Td[(1(t)A(t)S(t))]TJ /F5 11.955 Tf 11.96 0 Td[(S)]TJ /F7 7.97 Tf 6.59 0 Td[(1(t)_S(t)z(t) (3{45) Denethematrix,P(t)=S)]TJ /F7 7.97 Tf 6.58 0 Td[(1(t)A(t)S(t))]TJ /F5 11.955 Tf 12.42 0 Td[(S)]TJ /F7 7.97 Tf 6.59 0 Td[(1(t)_S(t),suchthatEquation 3{45 isexpressedasEquation 3{46 67
PAGE 68
_z(t)=P(t)z(t)(3{46) ThestabilityofthesysteminEquation 3{42 isequivalenttothestabilityofthesysteminEquation 3{46 ifthematrix,S(t)isaLyapunovtransformation.Suchequivalenceisparticularlyadvantageousinthatcertainconditionscanbeimposedsuchthatthestatematrix,P(t),inEquation 3{46 isuppertriangularforallt2R.Assuch,thepolesandassociatedstabilityofP(t)arerelativelyeasytoextract. ApolesetofP(t)isdenedbasedonitsnatureasuppertriangular.Thisset,asgiveninDenition 3.5.1 ,utilizestransitionmatricesforbothP(t)andS(t). Denition3.5.1. ThesetoffunctionsP:=fp1;:::;pnginL1(C)isapolesetofAifthereexistfunctionskij2CL1,i=1;:::;n;j>iandaninvertiblenxnmatrixS0suchthat S(t)=A(t;0)S0)]TJ /F7 7.97 Tf 6.59 0 Td[(1P(t;0);t2R+(3{47) isaLyapunovtransformationwhereP=fpijgand pij=8>>>><>>>>:pi:i=jkij:ij(3{48) AsuitablechoiceofS(t)hasbeenshowntobetheuniquesolutionofthematrixdierentialequation,giveninEquation 3{49 ,whichresultsinaLyapunovtransformation[ 126 127 ].NotethisexpressionagreeswiththepreviousdenitionforP(t)fromEquation 3{45 _S(t)=A(t)S(t))]TJ /F9 11.955 Tf 11.96 0 Td[(S(t)P(t);S(0)=S0(3{49) ThepolesetasgiveninDenition 3.5.1 dependsonthestate-transitionmatrix,A(t;0),associatedwiththesysteminEquation 3{42 .Thismatrixcanbedirectly 68
PAGE 69
computedusingknowledgeofA(t)soitrepresentsaknownquantity.Furthermore,thismatrixcanbeseparatedintoelementsusingastandardQR-decompositionusinganorthogonalmatrixofQ2RnnwithQQ1=IandanuppertriangularmatrixofR2RnnasinEquation 3{50 A(t;0)=Q(t)R(t)(3{50) TheelementsinEquation 3{50 arecomparedtotherelationshipinEquation 3{47 whichcanbeexpressedasA(t;0)=S(t)P(t;0)S)]TJ /F7 7.97 Tf 6.59 0 Td[(10.Inthisway,thecomputationofS(t)isdirectlycomputedastheorthogonalmatrixoftheq-rdecompositionwhilethetransitionmatrixassociatedwiththestabilitymatrixofP(t)iscomputedasthetriangularmatrixusingtheequivalanceinEquation 3{51 A(t;0)=[Q(t)][R(t)]=[S(t)]P(t;0)S)]TJ /F7 7.97 Tf 6.59 0 Td[(10(3{51) Inthisway,astraightforwardq-rdecompositionisusedtodetermineboththepolesofP(t)andthemodeshapesofS(t).Suchanapproachishighlyadvantageouscomputationallysincetheelementsareeasilyfoundusingmaturealgorithms. Anissueofnoteisthenon-uniquenatureofthisdecomposition.Essentially,thematricesassociatedwithaq-rdecompositioncanbescaledbyanyunitarymatrix.Theinformationrelatingtostabilityisretained;however,theneedtointerpretthepolesissomewhathighlightedbythisresult.Considerthatthisnon-uniquenatureimpliesthatthesignofanyrealvaluealongthediagonalofR(t)bearbitrarilychangedsotheconceptofamodemustreecttheabilitytovarythesignofP(t;0)usingtherelationshipfromEquation 3{51 3.5.2StabilityCriteria Anotionofstabilitymustbedenedforthetime-varyingsystemgiveninEquation 3{42 thatrelatesusefulproperties.Inthiscase,stabilityreferstothestateremainingbounded 69
PAGE 70
byanexponentialdecay.ThisnotionisuniformexponentialstabilityandisdescribedinDenition 3.5.2 Denition3.5.2. Thestateequation,_x(t)=A(t)x(t)withx(0)=xo,isuniformlyexponentiallystableifthereexistnite,positiveconstantsof;2RsuchthatkA(t;)ke)]TJ /F11 7.97 Tf 6.59 0 Td[((t)]TJ /F11 7.97 Tf 6.59 0 Td[() Asimilarnotionofstabilityisassociatedwiththepolesandtheirstabilitymodes.ThisnotionalsorequiresthetransitionmatrixtoremainboundedbyanexponentialdecayasnotedinDenition 3.5.3 [ 86 ]. Denition3.5.3. Thestabilitymodesassociatedwithp2CL1(C)isuniformlyexponentiallystableifthereexistsnite,positiveconstantsof;2Rsuchthatjp(t;t0)je)]TJ /F11 7.97 Tf 6.59 0 Td[((t)]TJ /F11 7.97 Tf 6.59 0 Td[(t0);8(tt00) Arelaxednotionofstabilityisalsodenedinwhichthetransitionremainsboundedandconvergestotheorigin.ThisnotionofasymptoticalstabilityisdescribedinDenition 3.5.4 [ 86 ]. Denition3.5.4. Thestabilitymodeassociatedwithp2CL1(C)isasymptoticallystableif,foranyto2R+,thereexistsanite,positiveconstantsuchthatjp(t;t0)j6;tt0andjp(t;t0)j!0;t!1. AnotionofsimplyremainingboundedisalsousedtodescribestabilityasuniformlystableinDenition 3.5.5 [ 86 ]. Denition3.5.5. Thestabilitymodeassociatedwithp2CL1(C)isuniformlystableifthereexistsanite,positiveconstantsuchthatjp(t;t0)j6;8(tt00). Finally,ageneralizedconceptofboundednessisgivenasnonexponentialinDenition 3.5.6 Denition3.5.6. Thestabilitymodesassociatedwithp2CL1(C)isnon-exponentialifthereexistsnite,positiveconstants1;2suchthat16jp(t;t0)j62;8(tt00). ThefundamentaltheoremisstatedinTheorem 3.5.7 whichnoteshowconditionsonthestabilitymodescanguaranteestabilityofthesystem.Theproofisnotpresentedhere;however,suchaproofisgivenintheliterature[ 86 ]. 70
PAGE 71
Theorem3.5.7. SupposethatA(t)hasapolesetofP.ThenthesystemofEquation 3{42 isuniformlyexponentiallystableifandonlyifpicorrespondstoanexponentiallystablemodefori2[1;n]. 3.5.3SpecialCases SpecialcasesdoexistwherethematrixfunctionA(t)isneitherconstantwithtimenorcontinuouslyvaryingwithtime.Infact,A(t)canfollowthenatureofaT-periodicsystemoronewithanitetime-variation. Recallingthefactthataperiodicstateequationcanequalatime-invariantstateequationbymeansofaFloquetdecomposition[ 39 ];Itcanbeshownthatthetime-invarianteigenvaluesdeterminethestabilityoftheequivalentperiodicstateequationandarecalledcharacteristicexponents[ 25 ].Thissetofcharacteristicexponentscanthenbesaidtoformapolesetoftheperiodicstateequation.Moreprecisely,supposethatAisT-periodic;thatis,thereexistsaniteT0suchthatA(t+T)=A(t)forallt2R.Thenitcanbeshown[ 86 ]thatthecharacteristicexponentsofAareapolesetofA. Whenthematrixfunction,A(t),hasanitetime-variation;thatisA(t)isconstantoutsideaniteinterval,itcanbesaidthatthefrozen-timeeigenvaluesformapolesetofA.Asmentionedearlier,thisisnotthecasewhenA(t)iscontinuouslyvaryingwithtime.Itisdirectlyduetothefactthatthefrozen-timeeigenvaluesmayormaynotcontaininformationabouttheoriginalstateequation,equation 3{42 .Moreprecisely,supposethatthereexistsaniteT>0suchthatA(t)=ATforalltT.Thenitcanbeshown[ 86 ]thesetoffrozen-timeeigenvaluesfi(t)gni=1satisfyingdet(A(t))]TJ /F5 11.955 Tf 12.53 0 Td[(i(t)I)=0ateacht2R+isapolesetofA. 3.6MethodologyComparisons Theeigenstructureforalineartime-varyingsystemcanbefoundbyvariousmethodologies,yeteachisspecictoitsown.Individualmethodsaredevelopedfromabasicsetofassumptionsspecictothemathematicalapproachitisbuiltupon,andthereforepossessesitsowncomparativeadvantages. 71
PAGE 72
Kamen'smethodinvolvescomputinganth-orderinput-outputdierenceequationwithtimevaryingcoecients.Thisdierenceequationisbasedonthesystem'slineartime-varyingdynamics,andfromit,anorderedpairofrightandlefteigenvaluescanbecalculated.Therighteigenvaluesarethenusedtodescribethesystem'stransientperformanceandstability.AfewbenetsofKamen'smethodinclude Arightpoleset,p21(t);p22(t),canbelinkedtothetime-invariantsystemviainitialconditions(frozentimeeigenvaluesatt=tdesired) TheVandermondematrix,V(t),diagonalizesthesystemmatrixwhenV(t)isinvertibleandbounded ThecalculatedLTVpolesareuniqueforalmostallinitialconditions Integratingtherightpolesetcaninuencethestabilityofthesystem whereas,thedisadvantagesinclude LTVpolescansometimeshaveanite-timesingularity,resultinginp21(t);p22(t)! Higherordersystemsaremorediculttoconvertintoaninput-outputordinarydierentialequation ZhuandJohnson'smethodalsoutilizesthenth-orderinput-outputdierenceequation,butinadditionaddsastatespaceapproachbyndingasimilaritytransformation,L(t),withdet(L(t))=constant,suchthat_z=L)]TJ /F7 7.97 Tf 6.59 0 Td[(1(AL)]TJ /F1 11.955 Tf 15.65 3.02 Td[(_L)z.InsteadofdeningLTVpolesbyrightandleftsets,ZhuandJohnsonuseorderedpolessetscalledSeriesD-spectra(SD-spectra)andParallelD-spectra(PD-spectra).AdeningbenetofZhuandJohnson'smethod,isthattheLTVpolesetscanbederivedaswell-dened(freeofnite-timesingularities). O'BrienandIglesias'smethodintroducesastate-spaceonlyapproachtosolvingforthetime-varyingeigenstructure.ItisshownthatbyapplyingaQRdecompositiontothecalculatedtransitionmatirx,eigenvectorsandeigenvaluescanbecomputed.BenetstoO'BrienandIglesias'smethodinclude Polesgiveinformationonstabilitythroughassociatedstabilitymodes 72
PAGE 73
Stabilitymodesarealwayscalculatedbeforepoles,thereforeassuringthatthepolesarealwaysbounded Moresuitableforhigherordersystems StabilityispreservedthroughLyapunovtransformations whereas,thedisadvantagesinclude TheLTVpolesblowuparoundii=0,wherepii=_ii.ii Thismethoddoesnotalwaysdiagonalizethesystem TheLTVpolescannotbelinkedtothetime-invariantsystem TheLTVpolesarenon-uniqueduetothenon-uniquenessoftheQRdecomposition Wu'smethodintroducesaslightydierentapproachtosolvingforlineartime-varyingsystemsinthestate-spaceform.X-eigenpairs,fi(t);ui(t)g,arefoundbyevaluatingthematrixdierential, 3{36 ,withanonsingularmatrix,S(t),developedfromtheeigenvectorsofA(t).BenetsofWu'smethodinclude Thesystemmatrixcanalwaysbediagonalized Thex-eigenpairinuencesstability whereas,thedisadvantagesinclude Thex-eigenpairisnotnecessarilylinkedtotheinitialLTIsystem Thex-eigenpairisnotunique 73
PAGE 74
CHAPTER4LINEARTIME-VARYINGMODALANALYSIS 4.1SystemDynamics Theconceptofaneigenvalue/eigenvectoranditsrelationshiptomodaldynamicsiswelldenedfortime-invariantsystems[ 88 ];however,thissamerelationshipissomewhatunderdevelopedfortime-varyingsystems.Thepurposeofthischapteristointroducenewconceptsusedfortime-varyingmodalanalysis.Tobegin,areviewofsomebasictime-invariantconceptswillbegiven. 4.1.1LinearTime-InvariantSystems:Formulation AsystemisconsideredtobeLinearTime-Invariant(LTI)ifitexhibitsbothlinearityandtimeinvariance. Linearitycanbedenedasanyrelationshipbetweenasystem'sinputandoutput,suchthatitpreservestheoperationsofsuperpositionandscalarmultiplication.Thepropertiesofscalarmultiplicationandsuperpositionofalinearsystem,y=f(x),areshowninEquation 4{1 ,respectively. f(x)=f(x)f(x1+x2)=f(x1)+f(x2)(4{1) Time-invariancealsoreferstoshift-invariance,whichstatesthatifaninputtoasystemisshiftedbysometime,t,thentheoutputofthatsystemwillalsobetimeshiftedbyt,asshowninEquation 4{2 y(t)=f(t)y(t)]TJ /F1 11.955 Tf 11.96 0 Td[(t)=f(t)]TJ /F1 11.955 Tf 11.96 0 Td[(t)(4{2) Inotherwords,atime-invariantsystemisonewhosebehavior(itsresponsetoinputs)doesnotchangewithtime. ThefundamentalresultinLTIsystemtheoryisthatanyLTIsystemcanbecharacterizedentirelybyasinglefunctioncalledthesystem'simpulseresponse.The 74
PAGE 75
outputofthesystemissimplytheconvolutionoftheinputtothesystemwiththesystem'simpulseresponse.Thismethodofanalysisisoftencalledthetimedomainpoint-of-view.Thesameresultistrueofdiscrete-timelinearshift-invariantsystemsinwhichsignalsarediscrete-timesamples,andconvolutionisdenedonsequences Equivalently,anyLTIsystemcanbecharacterizedinthefrequencydomainbythesystem'stransferfunction,whichistheLaplacetransformofthesystem'simpulseresponse(orZtransforminthecaseofdiscrete-timesystems).Asaresultofthepropertiesofthesetransforms,theoutputofthesysteminthefrequencydomainistheproductofthetransferfunctionandthetransformoftheinput.Consequently,convolutioninthetimedomainisequivalenttomultiplicationinthefrequencydomain MostLTIsystemconceptsaresimilarbetweenthecontinuous-timeanddiscrete-time(linearshift-invariant)cases. 4.1.2LinearTime-InvariantSystems:Response ConsideranLTIsystemdenedbyadierentialequation,asrepresentedinEquation 4{3 oritsequivalentstate-spacerepresentation,asrepresentedinEquation 4{4 xn+a1xn)]TJ /F7 7.97 Tf 6.59 0 Td[(1+:::+an)]TJ /F7 7.97 Tf 6.58 0 Td[(1_x+anx=p(t)(4{3) _x=Ax+Bu(4{4) Thecompletesolution,x(t),ofEquation 4{3 isderivedfromtwoparts,theparticularsolution(forcedresponse)andthecomplimentarysolution(naturalresponse).Theparticularsolutiondependsonthefunctionalformoftheinput,p(t),whilethecomplimentarysolutionisfoundbysettingtheinputfunctiontozeroandsolvingtheassociatedhomogeneousdierentialequation.Forthepurposesofthispaper,onlythecomplimentaryornaturalresponsewillbeconsidered. Foradynamicsystem,boththenaturalandforcedresponseconsistsofatransientandsteady-stateportion.Thetransientresponseisdenedastheresponsegeneratedby 75
PAGE 76
thesystem'schangefrominitialconditiontonalstate,whereasthesteady-stateresponseisdenedasthebehaviorofthesystem'soutputastimegoestoinnity. TheunforcedornaturalresponseforeachstatecanthenbedescribedasthesolutiontohomogeneoussolutiontoEquation 4{4 ,asseeninEquation 4{5 x(t)=nXi=1viexpit(4{5) TheeigenvectorisshowninEquation 4{5 asvi,whiletheeigenvaluesarerepresentedbyi.TakingthederivativeonbothsidesofEquation 4{5 andsubstitutingintheunforcedversionofEquation 4{4 for_x,resultsinthewellknowneigenvector-eigenvaluerelationshowninEquation 4{6 Ax=x(4{6) 4.1.3LinearTime-VaryingSystems:Formulation Lineartime-Varying(LTV)systemssharethesamelinearpropertiesasthepreviouslymentionedLTIsystem,yettheLTVsystemisnottime-invariantorshift-invariant.Thisfactsuggeststhatifaninputtoasystemisshiftedbysometime,t,thentheoutputofthatsystemwillnotbetimeshiftedbythesamet,asshowninEquation 4{7 y(t)]TJ /F1 11.955 Tf 11.96 0 Td[(t)6=f(t)]TJ /F1 11.955 Tf 11.96 0 Td[(t)(4{7) Assumingthatthelineartime-varyingstateequationisdenotedinstate-spaceform,asshowninEqutaion 4{8 ,andthatAisacontinuous,uniformlyboundedfunction(A2CL1),thenthestatespacerealizationcanbeshowntohavetheformdescribedbyEquation 4{9 _x(t)=A(t)x(t);x(0)=x0(4{8) 76
PAGE 77
G:=264A B C D375=8><>:_x(t)=A(t)x(t)+B(t)u(t)y(t)=C(t)x(t)+D(t)u(t)9>=>;(4{9) ItwillalsobeassumedthateachcoecientmatrixgeneratedbyG,isagainanelementofthecontinuous,uniformlyboundedspace(CL1). Forthecaseofmultivariablelinear,time-varyingsystems,aneigenstructureequation,reminiscentoftheclassicallydenedeigenvector-eigenvalueequationforconstantmatrices(Equation 4{6 ),canbedenedandshownasEquation 4{10 (A(t))]TJ /F5 11.955 Tf 11.96 0 Td[(pi(t)I)x(i)(t)=_x(i)(t);i=1;:::;n(4{10) InEquation 4{10 ,thetime-varyingeigenvalueisdenedaspi(t),wheretheassociatedtime-varyingeigenvectorisdenedasx(i)(t).Thestabilityofthestateequation,Equation 4{8 ,hasbeenshowntohavearelationdirectlylinkedtothebehaviorofthetermsinEquation 4{11 [ 126 ];[ 127 ]. Zt0pi()dx(i)(t);i=1;:::;n(4{11) Thestabilityofthestateequationhasalsobeenshowntohavearelationdirectlyrelatedtothemodesasscoiatedwitheachpolepi(t)[ 86 ];[ 130 ];[ 131 ],asshowninEquation 4{12 Zt0pi()d;i=1;:::;n(4{12) 4.1.4LinearTime-VaryingSystems:Response First,considerthetime-varyingresponseequationdenedusingthemodedenitionsprovidedbyWu[ 126 ]andKamen[ 59 ],asshowninEqaution 4{13 x(t)=nPi=1Cii(t)exphRt0pi()dix(t)=nPi=1Cii(t)(4{13) 77
PAGE 78
InEquation 4{13 ,thetime-varyingpoles,pi,arecomposedsuchthatpi:R+!C,whilethetime-varyingeigenvectors,i,arecomposedsuchthati:R+!CN.Itshouldbenotedthatthetime-varyingeigenvectorsareassumedtobedierentiableforalltime,tandthateacheachlinearly-independentsolution,i(t),canbedenedasamodeofthesystem[ 126 ]. Asecondtime-varyingresponseequationcanbedenedusingthemodedenitionsprovidedbyO'Brien[ 86 ],asshowninEqaution 4{14 x(t)=A(t;0)x0(4{14) InEquation 4{14 ,A(t;0)isrepresentativeofthetime-varyingstatetransitionmatrixandcanberelatedtothetime-varyingpolesandeigenvectorsviaEquation 3{51 BothEquation 4{13 and 4{13 mayberearrangedsuchthatarelationshipcanbedevelopedbetweenthetime-varyingpoles/eigenvectorsandthecoeeicentmatrix,A(t).ThisrelationshipisdeemedanLTVEigenrelation[ 112 ],anddependingonthetime-varyingalgorithmused,canbeshownaseitherEquation 3{29 3{32 ,or 3{49 .Anysetoftime-varyingpolesandeigenvectorsthatsatisfytheseequationswillbereferredtoasanLTVEigenpair[ 112 ]. GiventhisdenitionofanLTVeigenpair,itcanbeshownthatneitherthestabilitynorthefrequencycharacteristicsofanLTVsystemcanbedeterminedfromeithertheLTVpolesoreigenvectorsalone.Instead,eacheigenpairmustbeanalyzedtogetherasawhole.SinceLTIanalysistechniquesarebasedontheassumptionthatstabilityandoscillatorycharacteristicsareisolatedwithinthepoles,newanalysistechniquesareclearlynecessary. 4.2ModalInterpretation:Kamen Considertheoscillatoryresponseofa2-statesystem,asoriginallygiveninEquation 3{10 ,withthecoecientsnormalizedtoeasepresentationasinEquation 4{15 .SubstitutethedenitionofmodeintermsofpolesfromEquation 3{9 togenerateEquation 4{16 .Also, 78
PAGE 79
assumethegeneralizedpolestobecomplexconjugatessuchthatp21=p22=pR+|pIasinEquation 4{17 .ThecomplexexponentialscanthenbeexpressedintermsofsinesandcosinesasinEquation 4{18 andcombinedtogenerateEquation 4{19 x(t)=1(t)+2(t) (4{15) =eRt0p21()d+eRt0p22()d (4{16) =eRt0(pR()+|pI())d+eRt0(pR()+|pI())d (4{17) =eRt0pR()dcosZt0pI()d+|eRt0pR()dsinZt0pI()d+eRt0pR()dcosZt0pI()d)]TJ /F5 11.955 Tf 11.96 0 Td[(|eRt0pR()dsinZt0pI()d (4{18) =2eRt0pR()dcosZt0pI()d (4{19) AnoscillatoryresponsewithdecayisdemonstratedinEquation 4{19 tobegeneratedbyapairofpoleswhicharecomplexconjugates.Thenatureoftheoscillationsandthedecayarebothdeterminedbytheintegralsofrealandimaginarypartsofthesepoles.Also,theequal-but-oppositenatureoftheimaginarypartsofthesepolesmeanstheresponseinEquation 4{19 issimplydoubletherealpartofthemodeasnotedinEquation 4{18 4.2.1Damping Thedecayingnatureoftheresponse,whichissimilartothedampingratioofalineartime-invariantsystem,isdeterminedbythevaryingmagnitudeoftheexponentialinEquation 4{19 .TheresultingenvelopeisgiveninEquation 4{20 usingtherealpartofthepoleandequivalentlyinEquation 4{21 byaddingthecomplex-conjugatepoles. envelope(x(t))=eRt0pR()d (4{20) =eRt0p21()+p22() 2d (4{21) 79
PAGE 80
4.2.2Frequency Theoscillationsoccurwithafrequencyrelatedtotheimaginarypartofthetime-varyingpole.Atime-varyingequivalenttonaturalfrequency,!(t),isgeneratedbycomparingthecosinetermofEquation 4{19 toastandardtermindynamicsofcos(!t).Thetime-varyingequivalenttoanaturalfrequencyresultsfromthiscomparisonandisgiveninEquation 4{22 .Notethatanystablesystemwillhave!tendto0astimeincreases. !(t)=Rt0pI()d t(4{22) TheperiodicityoftheoscillationsresultsdirectlyfrominvertingthefrequencyofEquation 4{22 andisgiveninEquation 4{23 .Alternatively,suchperiodicitycanresultsimplybynotingwhentheangleinthecosinetermrepeatsitselfasgiveninEquation 4{24 T(t)=2t Rt0pI()d (4{23) =maxT>02R2 2T:Zt0pI()d=Zt+T0pI()d=0 (4{24) Also,atime-varyingequivalenttodampingratioiscomputedbyrelatingtheenvelopeofEquation 4{20 andthefrequencyinEquation 4{22 .Essentially,theenvelopeisequivalentto)]TJ /F5 11.955 Tf 9.3 0 Td[(!nandthefrequencyisrelatedto!np 1)]TJ /F5 11.955 Tf 11.95 0 Td[(2.theresultingdampingratioisgiveninEquation 4{25 (t)=vuut 1 1+Rt0pI()d Rt0pR()d2(4{25) 80
PAGE 81
4.3ModalInterpretation:O'Brien 4.3.1DampingandNaturalFrequency Considerthesame2-stateoscillatorysystemdescribedbyEquation 3{10 ,butnowrepresentedasthestate-spacesystemshowninEquation 4{4 Theresponseofthesystem'sdynamicswillvaryinmagnitudeaccordingtotheassociatedstability.Theparameterofdampingisusedtodescribetherateofvariationinthisresponsewithpositivedampingusedforstablesystemsthatdecaytozeroandnegativedampingusedforunstablesystemsthatgrowtoinnity. Therateofmagnitudevariationcanbeextractedfromthepairofpolesassociatedwiththemode.Considerthatthetime-invariantdynamicshavecomplex-conjugatepoleswith1=2=R+|IsoR=1+2 2.Asimilarrelationholdsforthepolesofthetime-varyingsystemasshowninEquation 4{26 envelope(x(t))=expp1(t)+p2(t) 2(4{26) Simulationresultssuggeststhatthetime-varyingfrequencyoftheoscillatorybehaviorcanbeapproximatedbythedominantzero-crossingperiodicityoftheeigenvectors.Forsimplicty,anapproximationisformulatedbyassumingthesystemresponsemaintainsasinusoidalnature.Inthiscase,thematricesderivedbydecomposingthestate-transitionmatrixwillalsohaveasinusoidalnature.Abasicpropertyofanysinusoidgivenasx(t)=sin(!t)hasx(t)=)]TJ /F5 11.955 Tf 9.3 0 Td[(!2sin(!t)sothefrequencyofoscillationisgeneratedby!=q )]TJ /F7 7.97 Tf 10.99 4.71 Td[(x x.Therefore,theresultingfrequencyofthetime-varyingsystemmaybeapproximatedbyEquation 4{27 !i(t)2r )]TJ /F5 11.955 Tf 10.49 8.09 Td[(qii qii(4{27) 81
PAGE 82
4.3.2ModeShapes Systemdynamicsoftenutilizemodeshapestodescribetheactualmotionofavehicle.Essentially,amodeshapepresentstherelativemagnitudebetweenstateswhichisindependentofanydecayoroscillations.Thesemodeshapesarenormallyfounddirectlyfromthetime-invariant,scalereigenvectorsderivedinEquation 4{6 .Unfortunately,thetime-varyingeigenvectorsfoundinthispaperdocontainfrequencyinformationandthereforearemorediculttointerpret. AmodeshapeisgeneratedbynotingtherelationshipbetweentheQandRmatrices;specically,considerthescalarvalues,rij(t),oftheupper-triangularmatrixgivenasR(t)andthecolumns,qi(t),oftheunitarymatrixgivenasQ(t). x(t)=nXj=1jXi=1qjrijxj()(4{28) Amodeshapecanbeimmediatelyrealizedfromtherstcolumn,q1(t)oftheQ(t)matrix.Thismodeshapeisidentiedbyconsideringthemotionthatwouldresultfromaninitialconditioninwhichtheonlynon-zerocomponentistherststate.Theresultingresponseissimplyascalar,r11(t)x1(),multipliedbythevectorofq1(t)asshowninEquation 4{29 .Themodeshapewhichdescribestherelativevariationsbetweeneachstateisthusgivenasv1(t)=q1(t). x(t)=q1(t)r11(t)x1()ifx()=[x1()0:::0] (4{29) =v1(t)r11(t)x1() (4{30) AnothermodeshapeisthendeterminedasacombinationoftherstandsecondcolumnsoftheQmatrix.Considertheresultingmotiongivenaninitialconditioninwhichonlythesecondstatehasanon-zerocomponent.ThemodeshapeinresponsetothismotionisactuallytheparentheticalexpressioninEquation 4{32 whichismultipliedbythescalarsofr22(t)andx2()togeneratetheresponse. 82
PAGE 83
x(t)=(q1r12+q2r22)x2()ifx()=[0x2()0:::0] (4{31) =q1(t)r12(t) r22(t)+q2(t)r22(t)x2() (4{32) =v2(t)r22(t)x2() (4{33) AgeneralizedexpressionforamodeshapeiscomputedbyextendingEquation 4{30 andEquation 4{33 tonth-ordersystems.SuchanexpressionisgiveninEquation 4{34 vi(t)=iXj=1qj(t)rji(t) rii(t)(4{34) Theresponseofthemorphingaircraftiswrittenintermsofthegeneralizedmodeshape,vi(t),fromEquation 4{34 .ThisresponseisalinearcombinationofthemodeshapesasshowninEquation 4{35 x(t)=nXi=1vi(t)rii(t)xi()(4{35) 4.4Example:SimpleMechanicalSystem Tohelpillustratetheconceptofdeterminingthelineartime-varyingpolesandstabilitymodes,usingaQRdecomposition[ 86 ],asimplesecondordersystemwillrstbeanalyzed.Inthisanalysis,correlationstolineartime-invariantsystemswillbemadeforthepurposeofcharacterizinglineartime-variantdynamicmodalproperties.Thisexampleiscomposedofamass-spring-dampersystem,asshowninFigure 4-1 ,andwillonlyconsiderthehomogeneousorunforcedsolution.Inaddition,anassumptionwillbemadethatthewheelsattachedtothecartarefrictionless. TheparametersforthisexamplearelistedinTable 4-1 ,whereitisseenthatonlythedampingcoecientchangeswithtime. Themassanddampingcoecientarechangedwithtime,asshowninEquation 4{36 ,insuchamannerthatthesystemexperiencesanincreaseinmassandspansallthree 83
PAGE 84
Figure4-1. Mass-Spring-DamperSystem Table4-1. Mass-spring-dampersystemparameters Parameterst=0st=20st=30s M(kg)279.5K(N/m)888C(Ns/m)0812011.5 dampingcharacteristics,simultaneously.Astimeprogresses,thesystemtransitionsfrominitiallybeingunderdampedandlighttocriticallydampedandmoderate,andnallytooverdampedandheavy.Thevaluesforthesetime-varyingmasses/dampingratiosandtheircorrelatedtimesareshowninTable 4-1 C(t)=0:4(t)M(t)=2+0:25(t)(4{36) StartingfrombasicprincipalsandNewton'ssecondlaw,asseeninEquation 4{37 ,theordinarydierentialequationdescribingthesystem'sdynamicscanbederived,asshowninEquation 4{40 .ItcanbeseenfromEquations 4{39 and 4{40 thatthetimerateofchangeinmass,_M(t),isrepresentativeofinertialeectsonthesystemanddirectlymodiesthecontributionofthetime-varyingdampingseeninthea1(t)coecient. 84
PAGE 85
F=d(Mv) dt (4{37) =Mdv dt+vdM dt (4{38) =Mx+_x_M (4{39) x+C(t)+_M(t) M(t)_x+K M(t)x=x+a1(t)_x+a0(t)x=0 (4{40) Fromobservation,itisdeterminedthatanequivalentstate-spacecanonicalrepresentationofEquation 4{40 canbeformedandshownasEquation 4{41 Ax+Bu=26401)]TJ /F5 11.955 Tf 9.3 0 Td[(a0(t))]TJ /F5 11.955 Tf 9.3 0 Td[(a1(t)375264x1x2375+264b1b23750(4{41) Thetwostatesofthesystem,representedinEquation 4{41 ,arethedisplacementposition,x,andtherateofchangeindisplacementposition,_x.Itisalsonotedthatthecoecientsfoundinthestate-spacestatematrix,a0(t)anda1(t),canberepresentedasfunctionsshowninEquations 4{42 and 4{43 a0(t)=8>>>><>>>>:0:t08 2+0:25t:0>>><>>>>:0:t00:2t+0:25t 2+0:25t:0
PAGE 86
Whenassumingthatinertialeectscanbeneglected,thenEquation 4{40 reducestotheEquation 4{44 andthestate-spacecoecient,a1(t),canberewrittenasshowninEquations 4{45 x+C(t) M(t)_x+K M(t)x=x+a1(t)_x+a0(t)x=0 (4{44) a1(t)=8>>>><>>>>:0:t00:2t 2+0:25t:0
PAGE 87
Thesystemresponseforeachcase,asshowninFigure 4-2 ,showthatthestatesinitallyoscillatebeforeeventuallyconverginguponasteadystateequilibriumvalue.Thisconvergenceoccursroughlyaroundtensecondsfortheinertialcaseandaround20secondsforthenon-inertialcase. A B Figure4-3. Time-VaryingModes:A)Non-Inertal:Mode1(|),Mode2()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()B)Inertial:Mode1(|),Mode2()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() ReferencingFigure 4-3 ,itiscleartoseethatthemodalresponses,forbothinertialandnon-inertialcases,alsoexhibitaperiodofdecayingoscillationsbeforenallyconverginguponzero.Notethattheconvergencewindowineachcaseisthesameasthatpreviouslyseeninthesystemresponses.Itshouldalsobenoticedthatbothmodesforeachcaseareboundedbysomeconstant,,andconvergetozeroastimegoestoinnity,thereforesuggestingthatthesystemmodesareasymptoticallystable[ 86 ].Animportantobservationtonoteisthatwhenthemodesapproachzero,thecorrespondingpolesbecomeincreasinglyill-conditioned.Theill-conditioningisbroughtaboutbythederivationofanLTVpole,asshowninSection 3.5 ,whichisdenedbydividingamode'stimerateofchangebythemodeitself.Therefore,whenthevalueofeachmodecrossesbelowsomepre-determinedthreshold,thepolevalueisceasedtobecalculated,asshowninFigures 4-4A and 4-4B 87
PAGE 88
A B Figure4-4. Time-VaryingPoles:A)Non-Inertal:Pole1()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[(),Pole2()-221()]TJ /F1 11.955 Tf 27.23 0 Td[(),RealportionofdiscreteLTIpoles(|),Poleaverage()B)Inertial:Pole1()-219()-219()]TJ /F1 11.955 Tf 33.14 0 Td[(),Pole2()-218()]TJ /F1 11.955 Tf 27.15 0 Td[(),RealportionofdiscreteLTIpoles(|),Poleaverage() Theassociatedpolesforthemass-spring-dampersystem,showninFigure 4-4A ,arefoundtooscillate,inanequal-but-oppositefashion,aboutalinedrawnthroughtheaverageofthetwoLTVpoles,(pole1+pole2)=2.Thispoleaverageisfoundtomatch,atdiscretepointsintime,therealpartofthecomplex-conjugateLTIpoles,asshowninFigure 4-4A .AnotableobservationisthatboththeLTIandLTVcomplex-conjugatepolessharethesamesimilarity,suchthattheaverageofthetwocomplex-conjugatepolesequaltherealpartsofeachLTIpole.ThisobservationsuggeststhatthemodaldampingforanLTVsystem,excludinginertialeects,isdirectlyrelatedtotheaverageofitsoscillatingpolepairsandcanbeapproximatedastheLTIdamping,asshownbyEquation 4{46 !LTV=(pole1+pole2)=2=Re(pLTI)(4{46) Thecorrespondingnon-inertialeigenvectors,asderivedinSection 4.3.2 ,canbeshowninFigure 4-5 Conversely,fromFigure 4-4B itisseenthatwheninertialeectsareincluded,theLTVpoleaveragedoesnotcorrespondwiththerealpartofthecomplex-conjugateLTI 88
PAGE 89
A B Figure4-5. Non-InertialTime-VaryingEigenvectors:A)Eigenvector1:(1,1)(|),(2,1)()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()B)Eigenvector2:(1,1)(|),(2,1)()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() poles.Instead,itisobservedthattheLTVpoleaveragediersfromtherealpartoftheLTIpolebyafactorroughlyequivalenttothemassrateofchange,_M.ThisfactsuggeststhatthedampingoftheLTVsystem,experiencingmassdisplacment,cannotaccuratelyberepresentedastherealpartsofeachLTIpole.Rather,thedampingoftheinertialLTVsystemcanberelatedtothediscreteLTIdampingbythesummationoftheLTVpoleaverageandthemassrateofchange,asshowninEquation 4{47 .ThecorrespondinginertialeigenvectorscanbeshowninFigure 4-6 !LTV=(pole1+pole2)=2=Re(pLTI))]TJ /F1 11.955 Tf 17.59 3.02 Td[(_M(4{47) Althoughsimilarintheory,theillustrativerepresentationofsuchacomplex-conjugatepolevariessignicantlyfromtheclassicallydenedLTIsystem[ 97 ]totherelativelynewdenedLTVsystem[ 86 ];[ 126 ].Thecomplex-conjugaterepresentationfortheaforementionedLTVsystemisshowninFigure 4-4A asanoppositephase,oscillatingpolepair.Frequencyinformation,however,isnotastrivialtondfromanoppositephase,oscillatingLTVpolepairasisthedamping.AsmentionedinSection 4.1.4 ,thefrequencyinformationissplitbetweenthepoleandeigenvector;therefore,neitherthe 89
PAGE 90
A B Figure4-6. InertialTime-VaryingEigenvectors:A)Eigenvector1:(1,1)(|),(2,1)()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()B)Eigenvector2:(1,1)(|),(2,1)()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() polenortheeigenvector,alone,canprovideacompleterepresentationofthetime-varyingfrequency.However,simulationresultssuggestthatasucientapproximationofthemodalfrequenciesmaybefoundbyreferringtotheperiodicityofthecolumnvectorsoftheorthonormalmatirx,Q,asshowninFigures 4-7 4-12 .AssumingthatQissinusoidalinnature,thenthisresultcanbeshownasthesamefrequencyapproximationderivedpreviouslyinEquation 4{24 A B Figure4-7. Non-InertialQcolumnvectorvalues:A)Q11(|),Q21()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()B)Q12(|),Q22()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() 90
PAGE 91
Figures 4-7A 4-7B representthecolumnsofthenon-inertial,time-varyingQmatrixwhichcorrespondtotheLTVpolesshowninFigure 4-4A .FromFigures 4-7A 4-7B ,itisseenthatthetermsQ11andQ22representthesameresponse,whilethetermQ12representsthesameresponseasQ21,justoutofphaseby180degreesorradians.Asaresult,theperiodicityforeachoftheseresponsesisidenticalandthereforesuggeststhatthemodalfrequencycanbefoundbyinvertingthemeasuredQcolumnperiodicityandmultiplyingby2,asshowninFigures 4-8 and 4-9 Figure4-8. Qcolumnperiodicity:Q11(|),Q21()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() Figure4-9. Qcolumnfrequency:ImaginarypartofdiscreteLTIpole(|),1Qcolumnperiodicity()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() 91
PAGE 92
Figures 4-10A 4-10B representtheinertialLTVQcolumnswhichcorrespondtotheLTVpolesshowninFigure 4-4B .ThesameconclusionscanbedrwanfromFigures 4-10A 4-10B ,asthatpreviouslydeterminedfromFigures 4-7A 4-7B .However,itshouldbenotedforthiscasethattheQtermsbecomeincreasinglyill-conditionedasthecorrespondingmodes,showninFigure 4-3B ,approachzero,asshowninFigures 4-6 and 4-10 .Thisconditionisadirectresultofthealgorithmusedtoanalyzethetime-varyingsystem. A B Figure4-10. InertialQcolumnvectorvalues:A)Q11(|),Q21()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()B)Q12(|),Q22()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() Aspreviouslyshown,themodalfrequencycanbefoundbyinvertingthemeasuredQcolumnperiodicityandmultiplyingby2.TheresultingQcolumnperiodictyandmodalfrequencyfortheinertialcasecanbeshowninFigures 4-11 and 4-12 ItcanbeseenfromFigures 4-9 and 4-12 thatthetime-varyingfrequencies(forboththeinertialandnon-inertialcases)aresimilartotheimaginarypartsofthetime-invariantpoles.Asaresult,itmaybeinferredforthisexample,thattheinclusionofinerialeectsdonotsignicantlyalterthefrequecyinformation;therefore,theimaginarypartsofthetime-invariantpolesmaybeusedtoapproximatethetime-varyingfrequency. 92
PAGE 93
Figure4-11. Qcolumnperiodicity:Q11(|),Q21()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() Figure4-12. Qcolumnfrequency:ImaginarypartofdiscreteLTIpole(|),LTVapproximatedfrequency()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() 93
PAGE 94
CHAPTER5CONTROLDESIGN 5.1InherentClosed-LoopNonlinearity Anymorphingsystem,regardlessofopen-looplinearity,willhaveaclosed-loopnonlinearityinthepresenceofstatefeedback[ 28 ].Theintroductionofthisnonlinearityhasbeenespeciallyproblematicforthecommunityinvestigatingactivesystems.Considerthatmorphingaircrafthavebeenextensivelystudiedbythecommunitybuthavefocusedpredominatelyonaerodynamicperformance[ 122 ]andmaterials[ 125 ].Inparticular,theseprogramshavedemonstratedclearbenetsofmorphingasmeasuredbyaerodynamicperformancerelatedtoliftanddrag[ 58 ].Aircrafthavebeenoptimized[ 18 ]usingmorphingtoimproveliftandminimizedragalongwithfuelconsumption[ 14 ],rangeandendurance[ 43 ],costandlogistics[ 15 ],actuatorenergy[ 93 ],andmaneuverability[ 99 ].Additionally,aeroelasticeectshavebeenoftenstudiedrelativetomaximumrollrate[ 66 44 9 ]andactuatorloads[ 77 ].Someinvestigationshaveconsideredcontrolofmorphingsystems;however,theydonotfullyaddressmanueveringight.Astudyusingpiezoelectricmaterialsdesignedcontrolonlytoachieveroll.[ 70 ]Avarietyofotherstudieshaveinvestigatedcontrolbutonlyinthecontextofacutationenergy[ 57 ]andcontrolauthority[ 19 ]forsimplechangesinightcondition.Otherstudieshaveconsideredmorphingforcontrolsurfacesbutnotassociatedcontrolsynthesis.[ 101 102 ]Also,astudydesignednonlinearcontrollersbutitsmorphingmodelusedsimplyadistributedsetofcontroleectorsratherthanashapewithfullytime-varyingdynamics.[ 117 ] Toillustratethenotionofthisinherentclosed-loopnonlinearity,rstconsidertheLIVgeneralizedmodel,asshowninEquation 5{1 .Thecorrespondingstatevectorisgivenasx2Rnwhilethecontroleectorisgivenas2R. _x=A()x+B()u(5{1) 94
PAGE 95
Assumingthereisnocontrolmatrix,B,andthatthestatedynamicshaveananedependence,usingA0;A1;A22Rnn,thenEquation 5{1 mayberewritteninthebilinearform[ 17 81 ],asshowninEquation 5{2 .Itshouldbenotedthatthemorphingrate,_,asseeninEquation 5{2 ,isgeneratedfromtheinclusionoftheinertialrates,_I,asdescribedinSection 2.3.1.2 _x=(A0+A1+A2_)x(5{2) Forthepurposeofsimplicity,themorphingrateterm,asshowninEquation 5{2 ,isassumedtobezero.Therefore,toachievegeneralizedfeedbackinthesystem,agenericcontrollaw,asdenedinEquation 5{3 ,isappliedtothenewlytruncatedformofEquation 5{2 =Kx(5{3) Asaresult,thegeneralizednonlinear(inthestates)expressioncanbedevelopedwhichdescribestheclosed-loopsystem,asseeninEquation 5{4 _x=(A0x+A1Kx2)(5{4) 5.2Quasi-StaticApproach 5.2.1Synthesis Quasi-staticvariationsareconsideredwhenthetimescalesofmorphingaremuchlessthanthetimescalesofmaneuvering.Suchquasi-staticmorphingisactuallythemostprevalentoftheschemesbeingadoptedforUAVandpilotedvehicles.ThevariablesweepontheF-14isanexampleofsuchaschemealongwiththevariablespanbeingproposedbyDARPA[ 123 ].Thequasi-staticmorphingisessentiallybeingusedtochangeightconguration,suchascruiseordive,relatedtosegmentsofthemissionprole.Theresultingdynamicsareslowlytime-varyingwithrespecttothemorphingduetothe 95
PAGE 96
disparity,whichmaybeordersofmagnitude,betweentimescalesofcongurationandmaneuvering. Thecontrollersneededtohandlethisquasi-staticmorphingwouldhavetobedesignedsuchthattheychosetheoptimalvalueofmorphingalongwithobtainingdesiredperformancemetricsformaneuvering.Ifitwereassumedthatanaircrafthadasetofslowactuatorstoalterightconguration,aswellas,asetoffastactuatorsformaneuvering,thenthefollowingactuationchoiceswouldensue.Thechoiceofmorphingactuationwouldbe,inessence,choosinganoptimaldesignfromwithinarangeadmittedbythemorphingscheme,whereasthechoiceofmaneuveringactuationwouldbeastandarddesignoftheautopilot. Asynthesismethodforthisapplicationwouldmostlikelybeformulatedasastandardmodel-followingapproachwhichallowedhandlingqualitiestobedirectlyutilized.Thecontrollerwouldseektocommandthemorphingsystem,P()=fA();B();C;Dg,torespondsimilarlytoadesiredsystem,X=f^A;^B;C;Dg,andtrackmaneuveringcommandsthroughouttheightenvelope.Therststepinthedesignprocesswouldneedtobethechoiceofmorphingvalueswhichminimizesthedierencebetweenthemorphingdynamicsandthedesireddynamics(forthemissionsegment),asinEquation 5{5 .Thenextstepwouldneedacontrollertocommandtheeectorsusedformaneuvering.ThiscontrollercouldbederivedfromtraditionalapproachesforrobustornonlinearsystemsasinEquation 5{6 =argmin2RW1kA())]TJ /F1 11.955 Tf 15.04 3.03 Td[(^Ak+W2kB())]TJ /F1 11.955 Tf 14.75 3.03 Td[(^Bk(5{5) K=argminK2SkX)]TJ /F5 11.955 Tf 11.96 0 Td[(Fl(P;K)k(5{6) 5.2.2Example:Gull-WingedAircraft Aninitialcontrollerhasbeensynthesizedforavariablegull-wingedaircraft[ 2 ].Asetofdesireddynamicswereformulatedforcruise-loiterandaggressive-maneuver 96
PAGE 97
r H() G P() 6 X() ? -e Figure5-1. Model-FollowingSystem congurationsbyvarying,forexample,frequencyoftheshort-periodmode.Thevalueofgull-wingmorphingisdeterminedusingEquation 5{5 toapproximatethesedesiredcharacteristics.AfeedbackcontrollerwasdesignedfromEquation 5{6 usinganinner-loopH1,G,andouter-loopPID,H,withgainschedulingacrossconguration,,asshowninFigure 5-1 .Apredenedmissionprolefromcruisetoaggressiveandbacktocruise,showninFigure 5-2 ,indicatesthatthemorphingvariedwithdesiredmodalpropertieswhiletheresponsestopitchdoubletstrackedthedesiredmaneuvers. Figure5-2. SimulationsoftheGull-MorphingAircraftModel Thisexampledemonstratestheeectivenessoftheapproach;however,severalissuesarenotyetrigorouslyaddressed.TherstissuewouldbedeningcomputationalstrategiesforEquation 5{5 whenvariousordersofpolynomials(A())wereusedwithmultipletypesofmorphing,.Asecondissuewouldinvolveconsiderationofndingastandardizedformulation,suchasLPV,wherethemaneuveringcontrollercouldinherentlyincludegainschedulingandrobustness. 5.3StabilizingControl:DisturbanceRejection 5.3.1Synthesis Aclassofsuboptimalcontrollersaresynthesizedfordisturbancerejection.Theabilitytorejectdisturbances,suchaswindgusts,isobviouslyanimportantfeatureofanyight 97
PAGE 98
controller.Theresultingclosed-loopsystemwillbeinherentlynonlinearduetotheeectsofstatefeedback;consequently,thecontrollerswillensuretheorigin,ortrimcondition,isgloballyasymptoticallystableforthesenonlineardynamics.[ 5 49 54 56 65 76 109 ] Anapproachfordisturbancerejectionhasbeenformulatedthatusesaconstant-gainmatrixforfeedback.ThisapproachguaranteesthedesiredstabilitypropertiesasstatedinLemma1.Essentially,theargumentresultsfromshowingthatalltrajectoriesoftheclosed-loopsystemareniteandconvergetozero. Lemma5.3.1. Giventhesystem,_x=(A0+A1)x,thentheoriginisgloballyasymptoticallystableusingthecontrollaw=Kxif 1. A0<0 2. A1K)]TJ /F5 11.955 Tf 16.14 0 Td[(A0<0 Proof:Theclosed-loopsystemisformulatedas_x=A0x+A1Kx2whichisobviouslynonlinear.Thesolutiontothisdierentialequationcanbeexpressedasx(t)=)]TJ /F11 7.97 Tf 6.58 0 Td[(A0 A1K)]TJ /F11 7.97 Tf 6.59 0 Td[(A0e)]TJ /F12 5.978 Tf 5.76 0 Td[(A0t.Inthiscase,thesteady-statevalueshowslimt!1x(t)=0ifA0<0soalltrajectoriesconvergetotheoriginifA0isnegativedenite.Asingularitywillalsoarisesuchthatlimt!1x(t)=1ifA1K)]TJ /F5 11.955 Tf 12.38 0 Td[(Aoe)]TJ /F11 7.97 Tf 6.59 0 Td[(A0t=0.Thissingularitycorrespondstot=)]TJ /F7 7.97 Tf 6.58 0 Td[(1 A0lnA1K A0whichhasasolutiont2Rwitht>0ifA0<0andK
PAGE 99
2. A1K1<0 Proof: Theclosed-loopsystemcanbeexpressedas_x=(A0+A1K0)x+(A1K2)x3.DeneaLyapunovfunctionasV(x)=1 2x2sothefollowingconditionsaresatised. 1. V(0)=0 2. V(x)0 3. _V(x)=(A0+A1K0)x2+(A1K2)x4 NowV(x)isavalidLyapunovfunction,andconsequentlytheoriginisgloballyasymptoticallystable,ifA0+A1K0<0andA1K2<0. 2 Animportantconsiderationistheformulationoftheopen-loopmodel.Specically,thesecontrollersassumetheLIVdynamicscanbeexpressedasarst-orderfunctionofthemorphinginput.Suchanassumptionwillrestrictthesystemswhichcanbeconsideredforcontrolsynthesis;however,therestrictionisnotnecessarilyunreasonable.Theexamplesinthispapercanbesatisfactorilymodeledasrst-orderfunctionsand,ifsecond-orderfunctionsareneeded,theLemmascanbeextendedtoconsiderhigherorders. Anotherconsiderationistheuseofcontroleectors.Thesedesignsassumethatshape,representedby,istheonlyeectorusedtorejectdisturbances.Theinclusionoftraditionaleectorscanbeconsideredbuttheinitialresultsassumemorphingistheonlyeectoravailableforcontrol. Also,thesecontrollersdonotguaranteeanypropertiesrelatedtoperformance.TheLemmasmerelyindicatethestabilityoftheoriginforthenonlinearclosed-loopsystem.Noinformationaboutthespeedatwhichresponsesconvergetotheorigincanbeprovided.Theseconditionsactuallyapplydirectlytoresponsesfrominititalconditionsandarenotevenformulatedtoconsiderotherperformancemetricssuchastrackingorrobustness.Assuch,theseresultsarelimitedinnaturetosimplestabilityforgustrejection. 99
PAGE 100
5.3.2Example:Chord-VaryingMorphing AnothermodeloftheMAVisgeneratedbyassumingapurevariationinchord.Thismodelindicatessomepropertiesoftheightdynamicsofachord-varyingmorphingwhilethespanandcamberarekeptconstant.Thechordforthismodelischosentorangefrom11.43cmto22.86cmwhilethespanremainsat60.96cm.TherepresentativemodelinTORNADO[ 80 ]resultedincongurationssuchasthoseshowninFig. 5-3 Figure5-3. Chord-VaryingMAVModel(m) TheLIVframeworkisagainusedtorepresenttheightdynamics.ThemodelinEq. 5{7 indicatesarst-orderttothestate-spacematricesfortheightdynamicsatasetofchordvalues.Theperturbationtochord,,rangesbetween0:05715m. _x=0BBBBBBB@266666664)]TJ /F1 11.955 Tf 9.3 0 Td[(0:075)]TJ /F1 11.955 Tf 9.3 0 Td[(0:1240)]TJ /F1 11.955 Tf 9.3 0 Td[(9:81)]TJ /F1 11.955 Tf 9.3 0 Td[(1:538)]TJ /F1 11.955 Tf 9.3 0 Td[(7:45612:0002:686)]TJ /F1 11.955 Tf 9.3 0 Td[(5:848)]TJ /F1 11.955 Tf 9.3 0 Td[(32:9340001:000377777775+266666664)]TJ /F1 11.955 Tf 9.3 0 Td[(0:330)]TJ /F1 11.955 Tf 9.3 0 Td[(2:09300)]TJ /F1 11.955 Tf 9.3 0 Td[(3:957)]TJ /F1 11.955 Tf 9.29 0 Td[(15:3310025:32495:187)]TJ /F1 11.955 Tf 9.3 0 Td[(160:16000003777777751CCCCCCCAx(5{7) ApairofcontrollersaredesignedusingLemma 5.3.1 andLemma 5.3.2 forthemodelinEq. 5{7 .ThismodelsatisestheconditionsoftheLemmasincludingstabilityfor=0.TheresultinggainsforthesecontrollersaregiveninTable 5-1 Table5-1. Gainsforchord-varyingaircraftdisturbancerejectioncontroller controlleralgorithm Lemma1=0)]TJ /F1 11.955 Tf 9.3 0 Td[(100:1xLemma2=[0]+0)]TJ /F1 11.955 Tf 9.3 0 Td[(10)]TJ /F1 11.955 Tf 9.3 0 Td[(0:5x2 100
PAGE 101
Responsestoagustfortheopen-loopsystemandclosed-loopsystemsareshowninFig. 5-4 .Theopen-loopsystemreturnstotrimtodemonstratetheexpectedstabilitywhenthechordisnotvariedfromthenominalcondition.Theclosed-loopsystems,usingLemma 5.3.1 andLemma 5.3.2 ,arealsostableandabletorejecttheinitialdisturbance. TheresponsesinFig. 5-4 indicatethatchord-varyingmorphingcanindeedbebenecialtorejectinggustssinceeachcontrollerreturnsthevehicletotrimfasterthantheopen-loopsystem.Theincreaseinrejectionisperhapsnotoverlysignicantbutatleastisclearlynoticeable.Thecontrollerchangesthechordwhichactuallyhastheeectofreducingthemomentofinertiaand,subsequently,increasingthestabilityderivatives.Astheactuationplotshows,thecontrollerreducesthechordtorejectthisdisturbancebutthencanquicklyincreasethechordtoreturntoaerodynamiceciency. Figure5-4. AngleofAttackResponseandAssociatedChordofAircraft TheabilityofthecontrollertoreturnthevehicletotrimisalsoevidentintheresponsesofallstatesasshowninFig. 5-5 .Eachstatereturnstoequilibriumalthoughtheratesofdecayintheoscillationsaftertheinitialdisturbanceclearlyvarybetweenresponses. Thebehaviorofthecontrollers,andconsequentlytheclosed-loopresponses,canbedeterminedfromFig. 5-5 inassociationwithTable 5-1 .Inparticular,thisassociationcanexplainthedierencesbetweenactuationresponsesinFig. 5-4 .ThechordvariationsusingthecontrollerofLemma 5.3.1 showanoscillatorynaturewhereasthechordvariations 101
PAGE 102
Figure5-5. ResponsestoChord-VaryingMorphing fromthecontrollerofLemma 5.3.2 settlestoaconstantsteady-statevalue.Thedierentbehaviorsiscausedbythelinearandquadraticnatureofthecontrollers.Essentially,asmallchordiscommandedbyscalingsmallvaluesofthestatesbutanevensmallerchordiscommandedbyscalingquadraticvaluesofthosesmallstates. 5.4FeedForwardOptimalControl 5.4.1State-ResponseTracking Morphingcanbeusedtotrackadesiredsignal,yd(t),forthesystemresponse.Theactualresponseisexpressedasy(t)=P((t))wheretheplantmodelisexplicitywrittenashavingtime-varyingdynamicsduetothemorphingcommand.Inthiscase,morphingisrestrictedtobetheonlycontroleector. Anoptimizationcomputesthemorphingcommandthataddressestracking.Thisoptimizationsearchesoverthemorphingtrajectorytominimizethetrackingerrorateveryinstanceintime,asshowninEquation 5{8 min(t)s Zt0kyd(t))]TJ /F5 11.955 Tf 11.95 0 Td[(P((t))k2dt(5{8) 102
PAGE 103
SuchanoptimizationiseectivelyafeedforwardschemeasdepictedinFigure 5-6 .Thecompensatoractuallygeneratesanoptimaltrajectoryforthemorphingbasedonano-lineminimization yd (t) P((t)) -y Figure5-6. ArchitectureforState-ResponseTracking Suchafeedforwardelementrepresentsacriticalaspectofmanynonlinearcontrolschemes.Themorphingdynamicscertainlyhaveanonlineardependencyontheinputcommandsoamulti-elementcontrollerislikelynecessary.Someaspectoffeedbackmaybeutilizedfornoiseandrobustness;however,designingthisfeedforwardcontrollerisaninitialsteptowardsamulti-elementcontroller. 5.4.2Pole-ResponseTracking Amorphingcommandcanalsobegeneratedfortrackingofapoleresponse.Denepd(t)asthedesiredlocationfortime-varyingpoles.Suchdesiredpolescandirectlyrelatetotraditionalparametersoffrequencyanddamping.Assuch,thetrackingofapoleresponseisanalogoustotrackingofmodalparameters. Thepole,p(t),isnotedasbeingrelatedtothediagonalelementsofthedecompositionofthestate-transitionmatrix.Specically,theithtermisgivenaspi(t)=r)]TJ /F7 7.97 Tf 6.58 0 Td[(1ii((t))d/dt(rii((t)))whereriiisthediagonaltermofR(t)asconstrainedbyQ(t)R(t)=A(t;0).Note,thecompletepoleformulationmaybeseeninSection 3.5 TheoptimizationtocomputethemorphingcommandisgiveninEquation 5{9 min(t)s Zt0kpd(t))]TJ /F5 11.955 Tf 11.95 0 Td[(rii((t))_rii((t))k2dt(5{9) Suchoptimizationisagainafeedforwardscheme.Theblockdiagramforthisscheme,asshowninFigure 5-7 ,diersfromFigure 5-6 byincludingseveralelementsrelatedtothepoles.Specically,ablockofQRisincludedtorepresenttheQRdecompositionandablockofisincludedtorepresentthecomputationofthestate-transitionmatrix. 103
PAGE 104
pd (t) (P((t))) QR s ? -r)]TJ /F7 7.97 Tf 6.59 0 Td[(1ii_rii Figure5-7. ArchitectureforPole-ResponseTracking 5.4.3Example:Mass-SpringSystem 5.4.3.1System Amorphingmass-springsystemiscontrolledtotrackdesiredstateresponsesandpolesignals.Thissystem,asshowninFigure 5-8 ,isasinglemassattachedtoawallthroughaspringanddamper. Figure5-8. Mass-Spring-DamperSystem Controlisactuatedthroughthedamperofthissystem.Essentially,thedampingisassumedtobeadjustableinordertoalterthesystemdynamics.TheequationofmotionisthusgiveninEquation 5{10 .Thesedynamicsareobviouslydependentonthedampingsocommandingadampingtrajectorywillrequireanalysisusingtime-varyingpoles. mx=kx)]TJ /F5 11.955 Tf 11.95 0 Td[((t)_x(5{10) Themassis8kgandthespringconstantis2N/m. 5.4.3.2MorphingBasis Theoptimalmorphing,thatischosentogenerateasystemresponseorpoleresponsethattracksadesiredresponse,isrestrictedtobeingathird-orderpolynomialoftime.AnymorphingtrajectoryisthusrestrictedtoliewithinthesetofallpolynomialswithrealcoecientsasnotedinEquation 5{11 104
PAGE 105
23Xi=0itii2R(5{11) Also,thecoecientsofthemorphingtrajectoryareboundedtopreventthesystemfromimmediatelybecomingoverdamped.Thehigher-ordertermsinthepolynomialhavethepotentialtodominateastimeincreases;consequently,thesetermsarerestrictedmorethanthelower-orderterms.Theexactvaluesoftheseboundsaredeterminedsuchthatanymorphingtrajectoryallowsthesystemtovarybetweenunderdampedandoverdamped.ThevalueofthedampingratiogiveninEquation 5{12 representstheboundarybetweenunderdampedandoverdamped. =(t) 2p km(5{12) Boundsonthecoecientsareinitiallychosentolimitthetimeatwhichthesystembecomescriticallydampedto7.5s.Suchatimeisarbitrarilychosenforthisexamplebutmayeasilybealteredtoconsiderdierenttypesofresponses.Giventhistimeconstraint,thecoecientsofEquation 5{11 mustsatisfytheexpressionsinEquation 5{13 .Theprocedureactuallyndsthecoecientssequentiallybyconsideringonlyonecoecienttobeanon-zerovalueatatime.Indongso,thesolutiondeterminesthemaximumcontributionneededbyeachmorphingcoecienttoachievethedesireddampingtransient. iti 2p km=Pi(7:5)i 8=1i=1;2;3(5{13) Also,aconstraintisintroducedsuchthat(t)>0forallvaluesoftime.Thisconstraintnotesthatthedesiredresponseisassumedtobeboundedandthusthesystemmusthavepositivedamping.Thecoecientsofeachpolynomialareadditionallylimitedtomaintainthispositivecondition. TheresultingboundsonthecoecientsaregiveninTable 5-2 105
PAGE 106
Table5-2. Boundsoncoecientofmorphingbasis CoecientUpperBoundLowerBound 04-411.07-1.0720.14-0.1430.02-0.02 5.4.3.3TrackingaSystemResponse:PolynomialMorphing Asetofdesiredresponsesaregeneratedusingpolynomialmorphing.Thesepolynomialsrangeasarst-order,second-orderandthird-orderfunctions.TheoptimizedvaluesofthemorphingcommandsaregeneratedforfeedforwardcontrolandshowninTable 5-3 .TheoptimizedmorphingisessentiallyidenticaltothedesiredmorphingwhichisanticipatedsincethedesiredmorphingcanbeexactlyduplicatedbythemorphingbasisofEquation 5{11 Table5-3. Optimalpolynomialmorphingtotrackdesiredpolynomialmorphing desiredactual 0:5+t0:5+t0:5t+0:076t20:5t+0:076t2)]TJ /F1 11.955 Tf 9.3 0 Td[(0:75t+0:1t2+0:02t3)]TJ /F1 11.955 Tf 9.3 0 Td[(0:75t+0:1t2+0:02t3 Theoptimizedstateresponses,muchliketheoptimizedmorphingtrajectories,areessentiallyidenticaltothedesiredstateresponses.TheseresponsesareshowninFigure 5-9 Anotherdesiredresponseisattemptedtobetrackedusingthepolynomialmorphing;however,thisdesiredresponseisassociatedwithamorphingcomposedofasinusoidaladdedtoapolynomial.ThisdesiredmorphingcannotbeduplicatedusingtherestrictedbasisofEquation 5{11 sosomeerrormustresultinthetracking.TheoptimizedpolynomialformorphingisgiveninTable 5-4 alongwiththisdesiredmorphing. Table5-4. Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing desiredactual 0:94t+sin(t)0:5343+1:0700t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:0812t2+0:0094t3 106
PAGE 107
A B C Figure5-9. DesiredResponse(|)andOptimalResponse()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()forA)First-OrderMorphingandB)Second-OrderMorphingandC)Third-OrderMorphing TheresultingstateresponseisshowninFigure 5-10 asreasonablyclosetothedesireresponse.Someerrorispresent;however,thedesiredmorphingisnotexcessivelyfaroutsidetherangeofthemorphingbasissothefeedforwardcontrolisnearlyabletoprovidethecorrecttracking. 5.4.3.4TrackingaSystemResponse:Piecewise-PolynomialMorphing Apiecewise-polynomialmorphingisconsideredtoexpandtherangeofthemorphingbasisseeninEquation 5{11 .Inthiscase,themorphingbasisisallowedtobearst-orderpolynomialwhosecoecientscanvaryatdiscretepointsintime.Anadditionalconstraint 107
PAGE 108
Figure5-10. DesiredResponse(|)andOptimalResponse()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()forSinusoidalMorphing ofcontinuityisimposedonthispiecewise-polynomialmorphingtoensurephysicalactuatorscouldpotentiallyimplementsuchacommand. Anoptimalvalueofpiecewise-polynomialmorphingisgeneratedtotrackadesiredresponsethatcontainsasinusoid.TheresultingcommandisgiveninTable 5-5 alongwiththemorphingassociatedwiththedesiredresponse. Table5-5. Optimalpiecewise-polynomialmorphingtotrackdesiredsinusoidalmorphing desiredactual 0:94t+sin(t)0:068+1:734t:0t1:521:959+0:399t:1:52
PAGE 109
Figure5-11. DesiredResponse(|)andOptimalResponse()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()forSinusoidalMorphing variousorders.Theoptimalmorphing,asgiveninTable 5-6 ,isabletoeectivelymatchthemorphingassociatedwiththedesiredpoles. Table5-6. Optimalpolynomialmorphingtotrackdesiredpolynomialmorphing desiredactual 1:07t1:07t1+0:75t+0:024t21+0:75t+0:024t2)]TJ /F1 11.955 Tf 9.3 0 Td[(0:2t+0:13t2+0:039t3)]TJ /F1 11.955 Tf 9.3 0 Td[(0:2t+0:13t2+0:039t3 Theoptimizedpolesignals,muchliketheiroptimizedmorphingtrajectories,areessentiallyidenticaltothedesiredpolesignals.Formodalreasons,thepolesignalswillbetruncatedat10seconds.ThesesignalsareshowninFigure 5-12 Anotherpoleresponseisattemptedtobetrackedusingthepolynomialmorphing.Thisdesiredresponseisagainassociatedwithamorphingcomposedofasinusoidaladdedtoapolynomial.Duetodampingconstraints,itremainsthesamedesiredsinusoidalmorphingasthatusedforthestateresponse.TheoptimizedpolynomialformorphingisgiveninTable 5-7 alongwiththisdesiredmorphing. Table5-7. Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing desiredactual 0:94t+sin(t)0:8532+0:7648t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:1248t2+0:0200t3 109
PAGE 110
A B C Figure5-12. DesiredResponse(|)andOptimalResponse()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()forA)First-OrderMorphingandB)Second-OrderMorphingandC)Third-OrderMorphing Althoughusingthesamedesiredmorphing,theoptimizedpolynomialinTable 5-7 isdierentfromthatinTable 5-4 .Thisresultissimplyduetotheoptimizationofadierentcostfunction. TheresultingpolesignalisshowninFigure 5-13 .Someerrorispresent;however,thefeedforwardcontrolisagainnearlyabletoprovidethecorrecttracking. 5.4.3.6TrackingaPoleResponse:Piecewise-PolynomialMorphing Thepiecewise-polynomialmorphingisusedtotrackapoleresponseassociatedwithasinusoidalmorphing.TheresultingmorphingandthedesiredtrajectoryaregiveninTable 5-8 110
PAGE 111
Figure5-13. DesiredResponse(|)andOptimalResponse()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()forSinusoidalMorphing Table5-8. Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing desiredactual 0:94t+sin(t)0:104+1:673t:0t1:522:182+0:269t:1:52
PAGE 112
5.5H1FeedbackControl:WithoutInertialEects Feedforwardcontrol,asshowninSection 5.4 ,isonemethodwhichcanbeusedforoptimaltracking,whileanotherisH1feedbackcontrol[ 21 23 55 61 100 129 ].H1controltypicallyachievesoptimalityfromderivingacontroller(robustorstabilizing)whichsolvessomemathematicaloptimizationproblem[ 29 45 48 ].Inthecaseoftracking,arobustcontrollerisusedandisdesignedsuchthatsomedesiredlevelofperformanceisobtained.Thisrobustcontrollermustbecapableofmaintainingperformancedespitetheinclusionofexogeneousinputsintothesystem,suchasnoise,disturbance,andreferencesignals.Whenallexogeneoussignalsareavailabletothesystem,theproblemisthenconsidera"fullinformation"H1problem[ 24 71 ],asshowninEquations 5{14 5{16 _x(t)=Ax(t)+B2w(t)+B1u(t) (5{14) z(t)=C1x(t)+D11w(t)+D12u(t) (5{15) y(t)=C2x(t)+D21w(t)+D22u(t) (5{16) Itshouldbenotedthatxrepresentsthesystemstatevariables,yisthesystemmeasuredoutput,zisthesystemcontroloutput,uisthesystemcontrol,andallstatematriceshaveproperdimensions.Itshouldbenotedhowever,thatthefullinformationassumptionrequiresthattheirbefullstate,disturbance,andcontrolfeedback,asdescribedbyEquation 5{17 C2=[I00]TD21=[0I0]TD22=[00I]T(5{17) TheexogeneousinputsinEquation 5{17 maybedescribedbyEquation 5{18 w(t)=[drn]T(5{18) 112
PAGE 113
Forthepurposesofthisstudy,the"fullinformation"standardwillbeappliedtothelineartime-varyingproblem.However,itisobviouswhentransformedintothebilinearform,Equation 5{1 cannotaccuratelybedescribedbythetime-invariantsystemdenedinEquations 5{14 5{16 .Therefore,Equations 5{14 5{16 mayberedenedforthebilineartime-invariantsystem[ 105 ],asshowninEquations 5{19 5{21 _x(t)=(A0+A1u(t))x(t)+B2w(t)+(B0+B1x(t))u(t) (5{19) z(t)=C1x(t)+D11w(t)+D12u(t) (5{20) y(t)=C2x(t)+D21w(t)+D22u(t) (5{21) Thisnewbilineartime-invariantsystemisbaseduponEquation 5{2 withouttheinertialcontribution,asshowninEquation 5{22 _x=A0x+A1xu(5{22) Itshouldbenotedthatthelineartime-invariantsystem,showninEquation 5{1 ,wasassumedtohaveananedependencewithinthestates.Fromthisresult,itwasfurtherassumedthatthelineartime-varyingsystemhadananedependencewithintheinputcommand,u(t),thusgivingtheformofabilinearsystem.Inmakingthisassumption,itisseenthatthetime-varyingdependencehasmovedoutsideofthestatematrixintothecontrolinput,u,andstatevariable,x.Thisfactsuggeststhatabilineartime-invariantcontrollermaybeusedtosucientlycontroltheunderlyingtime-varyingsystem[ 4 68 ]. InordertosimplifytheH1bilinearproblemandprovideareasonablesolution,thesimplifyingassumptions,showninEquation 5{23 ,needtobemade. 113
PAGE 114
DT12C1=0DT12D12=IB0DT12=0D21DT21=ID11sucientlysmall(5{23) FromEquation 5{23 ,itismentionedthatD11mustbesucientlysmall.ThissuciencysuggeststhatthemaximalsingularvalueofD11mustbesmallerthanthecorrespondingH1normoftheclosed-loopcontrolsystem.ItisalsoseenfromEquation 5{23 thatA0andB0mustbestabilizable;However,itisrecalledthatforthetime-varyingprobleminquestion,thereexistsnoexternalinputindependentofthestatecommand.Asaresult,ithasbeenshown[ 106 ]thattheoptimalcontrolofabilinearsystemcanbeachievedifandonlyifthereexistsascalar,r,suchthattherealpartsoftheeigenvaluesof(A0)]TJ /F5 11.955 Tf 12.77 0 Td[(r(A1+B1))arenegativeandthecostfunction,asseeninEquation 5{24 ,isminimized. J=1 2Z10fzT(t)z(t))]TJ /F5 11.955 Tf 11.95 0 Td[(2wT(t)w(t)gdt(5{24) BycombiningEquation 5{24 withthepartialderivativeoftheclosed-loopLyapunovfunction,,theHamiltonianfunctioncanbedenedandwrittenasEquation 5{25 H=1 2zTz)]TJ /F5 11.955 Tf 13.15 8.09 Td[(2 2wTw+T_x(5{25) SubstitutingEquations 5{19 5{21 intoEquation 5{25 resultsinEquation 5{26 ,fromwhichthemaximalexogeneousinputvectorandminimalcontrolcommandmaybefound. 114
PAGE 115
H=1 2(xTCT1C1x+xTCT1D11w+xTCT1D12(u+r)+(u+r)TDT12D11w+wTDT11D11w+(u+r)TDT12D12(u+r))-222(2wTw)+T(Ax+(A1+B1)x(u+r)+B2w) (5{26) Themaximalexogeneousinputvector,w,andminimalcontrolcommand,u,arefoundbytakingthepartialderivativeofEquation 5{26 ,respectively,andsettingitequaltozero.TheresultingoptimaltermscanbedescribedbyEquations 5{27 5{28 u=)]TJ /F5 11.955 Tf 9.3 0 Td[(r)]TJ /F1 11.955 Tf 15.95 8.09 Td[(1 2(D12D11BT2)]TJ /F5 11.955 Tf 11.95 0 Td[(xT(A1+B1)T)+xT(A1+B1)T(5{27) w=)]TJ /F1 11.955 Tf 13.3 8.09 Td[(1 2(DT11D12xT(A1+B1)T)]TJ /F5 11.955 Tf 11.96 0 Td[(BT2)(5{28) UsingthesimplifyingassumptionsfoundinEquation 5{23 ,theclosed-loopformoftheHamiltonianmaybefoundbysubstitutingEquations 5{27 5{28 backintoEquation 5{25 .Theresultingclosed-loopformoftheHamiltoniancanbeshownbyEquation 5{29 HCL=1 2(xTCT1C1x+(A0)]TJ /F5 11.955 Tf 11.95 0 Td[(r(A1+B1))+(A0)]TJ /F5 11.955 Tf 11.95 0 Td[(r(A1+B1))TT)-222((A1+B1)xxT(A1+B1)TT+1 2((A1+B1)xxT(A1+B1)TT+B2BT2T)]TJ /F1 11.955 Tf 12.44 3.16 Td[((A1+B1)xDT12D11BT2)]TJ /F1 11.955 Tf 12.45 3.16 Td[(B2DT11D12xT(A1+B1))) (5{29) Theclosed-loopLyapunovfunction,,forbilinearsystemshasbeenshowntobeaquadraticfunctionofthestatevector[ 10 62 69 79 32 ]andthereforecantaketheformshowninEquation 5{30 =1 2xTPx(5{30) 115
PAGE 116
ThepartialderivativeofEquation 5{30 canbetakenwithrespecttoxtond,asshowninEquation 5{31 =@ @x=xTP(5{31) SubstitutingEquation 5{31 intoEquation 5{29 givesthenalversionoftheclosed-loopHamiltonian,asshownbyEquation 5{32 HCL=1 2xT(CT1C1+P(A0)]TJ /F5 11.955 Tf 11.95 0 Td[(r(A1+B1))+(A0)]TJ /F5 11.955 Tf 11.96 0 Td[(r(A1+B1))TP)-222(P(A1+B1)xxT(A1+B1)P+1 2(P(A1+B1)xxT(A1+B1)P+PB2BT2P)]TJ /F5 11.955 Tf 11.96 0 Td[(P(A1+B1)xDT12D11BT2P)]TJ /F5 11.955 Tf 11.95 0 Td[(PB2DT11D12xT(A1+B1)TP))x (5{32) IthasbeenshownforbilinearsystemsthatthealgebraicRiccatiequationisderivedfromthelinearpartsoftheclosed-loopHamiltonianfunction[ 106 ].Therefore,thealgebraicRiccatiequationrelatedtoEquation 5{32 canbeshownbyEquation 5{33 [ 92 ]. P(A0)]TJ /F5 11.955 Tf 11.95 0 Td[(r(A1+B1))+(A0)]TJ /F5 11.955 Tf 11.95 0 Td[(r(A1+B1))TP+CT1C1+1 2PB2BT2P=0(5{33) ForthesingleinputbilinearsystemdescribedbyEquation 5{22 ,itcanbesaidthattheclosed-loopnonlinearsystemsfromtheH1statefeedbackcontrolaregloballyasymptoticallystableifandonlyifthefollowingconditionsaremet: (A0;B2)stabilizable (A0;C1)observable eig(A0)]TJ /F5 11.955 Tf 11.95 0 Td[(r(A1+B1))<0 Itshouldbenotedthattheconstantvalue,r,canbefoundusingmethodssuchastheRouthcriterion. 116
PAGE 117
5.6FutureWorkandChallenges Multipleavenueshavebeenestablishedtocontinuethecontrolworkpresentedthroughoutthispaper.Twoareasofparticularinterestwouldbethethoroughinvestigationandderivationofadynamiccontrolapproach,aswellas,theinclusionofaninertialtermwithinboththefeedforwardandH1feedbackoptimaizations. Fromthesetwotopics,theareaclosesttorealizationatthistimewouldbetheinclusionoftheinertialterm,_,asshowninEquation 5{2 .Therefore,themainfocusofthissectionwillprimarilycenteraroundtheconceptsandchallangessurroundingtheintroductionofthisterm. ItisclearlyseenfromSection 2.3.1.2 thatinertiaplaysacrucialroleinthedynamicoutcomeofthedescribedtime-varyingsystem.Typically,inertialeectsareneglectedduetotheirsmallinuenceonthedynamics;However,theeectsofinertiaintime-varyingsystemsarestillyettobefullyunderstood,althoughintialtestingsuggeststhatslightchangesininertiacouldbesignicantifunaccountedfor[ 47 ]. Anobviouschallengeregardingtheintroductionoftheinertialterm,_,intotheoptimalcontrolarchitectureisthedecisionofwhichtermtakesprecedence,themorphingangleorthemorphingrate?Thesetwotermsarelinkedthrougharstderivativerelationshipandthereforecannotbecontrolledindependently.Asaresult,theoptimizationproblembecomeseitherndingthecorrectratewhichproducesthedesiredmorphingtrajectoryorviceversa.Thekeytothisconceptisnotjustndingthecorrectterm,butratherndingthecorrecttermwhichminimizestheeectoftheotherandmaintainsthedesiredobjective. Beforethisproblemcanbeaddressed,theabilitytoeectlystabilizethesystemshouldrstbeconsidered.ItwasshowninSection 5.3.1 thatatime-varyingsystem,withoutinertia,couldbesuboptimallycontrolledwhileguaranteeinggloballyasymptoticstability.Anaddendumtothisillustrationwouldbetoaddtheinertialterm,asshowninEquation 5{2 ,andrederiveasubotpimalcontroller. 117
PAGE 118
StartingwithEquation 5{2 ,itisseenbyredeningthemorphingrate,_,astheproductoftwopartialderivatives,thatEquation 5{2 canberedenedintermsof_x,asshowninEquation 5{34 _x=A0x+A1x+A2x@ @x@x @t=A0x+A1x+A2x@ @x_x(5{34) ItshouldbenotedthatsinceisscalarandEquation 5{34 representsaSISOsystem,thenEquation 5{34 maybewrittenassowithoutalgebraicissues. Afterrearrangingterms,Equation 5{34 canberewrittenintheformshowninEquation 5{35 _x=(I)]TJ /F5 11.955 Tf 15.86 0 Td[(A2x@ @x))]TJ /F7 7.97 Tf 6.59 0 Td[(1(A0x+A1x)=A)]TJ /F7 7.97 Tf 6.59 0 Td[(1~A(5{35) NoticingtherightsideofEquation 5{35 ,itisrecalledthat~Ahasalreadybeenshowntoachievegloballyasymptoticstabilityusingsuboptimalcontrollers.Therefore,Awouldneedtobegloballyasymptoticallystablefortheenitiresystemtoachievegloballyasymptoticstabilityoratleastgloballystableforthesystemtobestable. AnotherwaytolookattheproblemistoseethatEquation 5{35 mayberearrangedsuchthatittakestheformshowninEquation 5{36 _x=(I)]TJ /F5 11.955 Tf 15.85 0 Td[(A2x@ @x))]TJ /F7 7.97 Tf 6.59 0 Td[(1(A0+A1)x=A)]TJ /F7 7.97 Tf 6.59 0 Td[(1^Ax(5{36) Now,thequadraticLyapunovfunction,asshowninEquation 5{30 ,canbeusedtochecktheoverallstabilityofthenewlydenedtime-varyingsystem,asdescribedbyEquation 5{36 .SubstitutionofEquation 5{36 intothepartialderivativeofEquation 5{30 resultsinthestabilitymetricshowninEquation 5{37 @ @x=xT(I)]TJ /F5 11.955 Tf 15.86 0 Td[(A2x@ @x))]TJ /F7 7.97 Tf 6.58 0 Td[(1(A0+A1)x=xTAx(5{37) 118
PAGE 119
ItiscleartoseefromEquation 5{37 thatinorderforthetime-varyingsystem(withinertia)toachievestability,thenAwouldneedtobenegativedenite. 119
PAGE 120
CHAPTER6EXAMPLEOFVARIABLESWEEPAIRCRAFT 6.1Design 6.1.1BiologicalInspiration Theseagullisalogicalchoicefromwhichtoderivebiologicalinspirationsinceitissoadeptatagileyinginwindyconditions.Suchbirdsareroutinelyseentrackingboats,divingtocatchprey,andlandingonbuoysdespiteheavywindsandstronggustsfromdierentdirections.Themissionsenvisionedforaminiatureairvehiclerequireasimilarsetofabilities;therefore,abiomimeticapproachiswarranted. Theskeletalstructureoftheseagullisacriticalcomponentthatenablesightcapability.Inparticular,thejointsattheshoulderandelbowareusedtorotatethewingsandconsequentlyaltertheightdynamics.Suchrotation,asseeninFig. 6-1 ,causesdisplacementinbothverticalandhorizontaldirectionswhichcorrelatestowingdihedralandwingsweep. Figure6-1. PicturesofSeagulls Thewings,asshowninFig. 6-1 ,willusuallyvarythesweepbetweentheinboardandoutboard.Thevariationresultsfromtheindependentactuationabouttheshoulderandelbowtovarythehorizontalrotation.Thismorphingprovidesavarietyofchangesintheightcharacteristicssuchasstability,divespeed,andturnradius. Also,thewingsareshowninFig. 6-1 tovarythesweepbetweenrightandleftwingsalongwiththeinboardandoutboard.Thisvariationutilizes4degreesoffreedomresulting 120
PAGE 121
fromindependentactuationofshoulderandelbowoneachwing.Thismorphingenablesseveralmaneuversrelatedtohoming,rolling,andrejectingcrosswinds. Emphasisisplacedontherelationshipbetweenwingsweepandmaneuvers.Thesweepisalreadyadesignparameterwhoseeectsonaerodynamicshavebeenstudiedfortraditionalaircraft;however,thestudyofbirdsprovidesadditionalinsightintotheperformancethatmaybeachievableusingindependentmulti-jointsweep.Inthiscase,thecorrelationsbetweensweepanddiveareaugmentedwithcorrelationsbetweensweepandagilityforbothturningandtrimming. 6.1.2MechanicalDesign Avehiclewhichfeaturestheindependentmulti-jointcapabilityisdesignedbyretrottinganexistingaircraft[ 3 ].Thebasicconstructionusesskeletalmembersofaprepregnated,bi-directionalcarbonberweavealongwithrip-stopnylon.Thefuselageandwingsareentirelyconstructedoftheweavewhilethetailfeaturescarbonsparscoveredwithnylon.Theresultingstructureisdurablebutlightweight. Thewingsactuallyconsistofseparatesectionswhichareconnectedtothefuselageandeachotherthroughasystemofsparsandjoints.Thesejoints,asshowninFig. 6-2 ,arerepresentativeofashoulderandelbowwhichservetovarythesweepofinboardandoutboard.Therangeofhorizontalmotionadmittedbythesejointsisapproximately30deg. Figure6-2. JointsonWing 121
PAGE 122
Itisnotedthatconventionalaileroncontrolsurfacesareomittedfromtheaircraft'snaldesign.Thisfeatureisadirectresultofspan-wiseinconsistenciescreatedbythedynamicrangeofmorphingcongurations.Therefore,theelbowjointsaredesignedinsuchamannerthattheyallowbothhorizontalsweepandrollingtwist.Thismotionisaccomplishedbycreatingaoatingjointthatcloselymimicsthevariousrangesofmotionattainablebyanautomobile'suniversaljoint,asshownifFig. 6-3 Figure6-3. FloatingElbowJoint Thewingsurfacemustbekeptcontinuousforanycongurationofsweepingbecauseofaerodynamicconcerns.Thisvehicleensuressuchcontinuitybylayeringfeather-likestructures,asshowninFig. 6-4 ,withinthejoint. Figure6-4. Feather-LikeElements Thesestructuresretractontoeachotherunderthewingwhenboththeinboardandoutboardaresweptback.Conversely,theycreateafan-likecoveracrosstheensuinggap 122
PAGE 123
whentheinboardissweptbackandtheoutboardissweptforward.Thecontractionandexpansionofthesurfaceareacreatedbythesestructuresissmoothlymaintainedbyatractandrunnersystemimplementedontheouterregionsofeachmember,asseeninFigure 6-5 Figure6-5. Trackandrunnersystem Spars,formedfromhollowshaftsofcarbonber,areplacedalongtheleadingedgeofeachwing.Thesesparsactasbotharigidsourcetomaintaintheleading-edgecurvatureandaconnectionofeachindependentwingjoint.Theinboardsparistranslatedhorizontallybyaservo-drivenlinearactuatorlocatedinsidethefuselage.Theinboardsparisthenconnectedtotheinboardwingsectionattheshoulderjointlocatedontheoutsideofthefuselage.Theinboardsparthenconnectsattheelbowjointtooutboardsparatroughlythequarter-spanpoint.Theoutboardwingregionisactivatedindependentlyoftheinboardregionbymeansofaservoattachedattheelbow.Anillustrationofthesparconguration,withcorrespondingattachmentpoints,canbeseeninFig 6-6 6.1.3TechnicalSpecications Thevehicleisxed-wingdesignthatincludesafuselageandempennageforatotalweightof596g.Thefuselagehasatotallengthof48cmandiscomposedofacylindicalbaywithlargestdiameterof7cmthatstretchesfor30cmandaboomwithdiamterof1cmthatstretchesfor18cm.Thewingsaremountedonthetopofthecylindrical 123
PAGE 124
Figure6-6. Underwingsparstructure baywhiletheavionicsaremountedinsidethecylindricalbay.Thecomponentsoftheempennage,includingtheelevatorandrudderwhichactascontrolsurfaces,aredescribedinTable 6-1 Table6-1. Empennagespecications SectionReferenceReferenceReferenceSpan(cm)Chord(cm)Area(cm2) HorizontalTail30.487.62232.26VerticalTail15.247.62116.13Rudder30.484.57139.29Elevator15.243.0446.33 Referenceparametersforthemorphingwingvarybasedonsweepconguration.ArepresentativesetoftheseparametersaregiveninTable 6-2 foralimitedsetofsymmetriccongurationsinwhichtheleftandrightwingshaveidenticalsweep. Theavionicsconsistsofactuatorsandamotor.AsetofeightHitechHS-65MGmetalgearservosprovideactuationandaredistributedasthreeservostocontroltheinboardsweep,outboardsweepandwingtwistofeachwingalongwithtwoservostocontrolthe 124
PAGE 125
Table6-2. Referenceparametersforsymmetricsweep InboardOutboardReferenceReferenceReference(deg)(deg)Span(cm)Chord(cm)Area(cm2) -15-3066.1714.681028.11-10-2073.9713.121003.45-5-1078.8112.38976.250080.3911.84947.1151078.6111.62916.68102073.6111.69885.66153065.7212.13854.76 elevatorandrudder.ThemotorisanE-ightsix-seriesbrushlesselectricmotorpoweredbyathree-cellThunderPowerLi-polymer2100mAhbattery. 6.2Modeling 6.2.1ComputationalTools Aerodynamicsolutionsforthree-dimensionalwingsofanyshapeorsizecanbecalculatedbyusingavortex-latticemodel.Assumingtheowtobeincompressibleandinviscid,thewingismodeledasasetofliftingpanelswitheachcontainingasinglehorse-shoevortex.Bothspan-wiseandchord-wisevariationinliftcanbemodeledasasetofstepchangesfromonepaneltothenext,asshowninFig 6-7 Figure6-7. Modelingoftheliftvectors Locatedatthepanelquarter-chordpositionisaboundvortex,whichshedstwotrailingvortexlines.Therequiredstrengthoftheboundvortexoneachpanelwillneedtobecalculatedbyapplyingasurface-owboundarycondition.Thisboundaryconditionstatesthereiszeroownormaltothesurfaceofthewing.Foreachpanelthiscondition 125
PAGE 126
isappliedatthethree-quarter-chordpositionalongthecenterlineofthepanel.Thenormalvelocityismadeupofafreestreamcomponentandaninducedowcomponent.Thisinducedcomponentisafunctionofstrengthsofallvortexpanelsonthewing.Thus,foreachpanelanequationcanbesetupwhichisalinearcombinationoftheeectivestrengthsproducedfromallpanel.Bysolvingtheseequations,onecanproduceamodelthateectivelydescribestheaerodynamicqualitiesandcontrollabilityofanaircraft. AVL[ 30 ]isavortex-latticemodelthatisbestsuitedforaerodynamiccongurationswhichconsistmainlyofthinliftingsurfacesatsmallanglesofattackandsideslip.Thesesurfacesandtheirtrailingwakesarerepresentedassingle-layervortexsheets,discreditedintohorseshoevortexlaments,whosetrailinglegsareassumedtobeparalleltothelongitudinalx-axis,asshowninFig 6-8 Figure6-8. Modelingofthetrailinglegvectors AVLprovidesthecapabilitytoalsomodelslenderbodiessuchasfuselagesandnacellesviasource-doubletlaments.Theresultingforceandmomentpredictionsareconsistentwithslender-bodytheory,buttheaerodynamicsaregenerallychallengingtocompute,thereforethemodelingofbodiesshouldbedonewithcaution.Ifafuselageisexpectedtohavelittleinuenceontheaerodynamicloads,itshouldbeleftoutoftheAVLmodelentirely.Thisexclusionofthebodyisprescribedtoavoidpotentialinaccuraciesfromenteringtheoverallmodel. 126
PAGE 127
AVLassumesquasi-steadyow,whichallowsunsteadyvorticitysheddingtobeneglected.Moreprecisely,itassumesthelimitofsmallreducedfrequency,whichmeansthatanyoscillatorymotion(e.g.inpitch)mustbeslowenoughsothattheperiodofoscillationismuchlongerthanthetimeittakestheowtotraverseanairfoilchord.Thisassumptionisvalidforvirtuallyanyexpectedightmaneuver.Also,theroll,pitch,andyawratesusedinthecomputationsmustbeslowenoughsothattheresultingrelativeowanglesaresmall,asjudgedbythedimensionlessrotationrateparameters. 6.2.2SweepDetermination Thisvehicleisabletoachieveawiderangeofsweeporientationsinbothsymmetricandasymmetriccongurations.SomerepresentativecongurationsareshowninFig. 6-9 todemonstratetherange. Figure6-9. SweepCongurations Acoordinatesystemisdenedtofacilitatetheproperdescriptionofeachconguration.Sweepanglesassociatedwiththeinboardsectionsaredenotedas1fortherightwingand3fortheleftwing,whileoutboardsectionsuse2fortherightwingand4fortheleftwing.Theseangles,asshowninFig. 6-10 ,aredenedsuchthatpositivevaluesindicateabackwardsweep.Also,eachangleisdescribedrelativetotheright-sidereferencelinethatisperpendiculartothefuselagereference. 127
PAGE 128
6nose -rightside 1 2 3 4 Figure6-10. SweepAngles 6.3AerodynamicProperties 6.3.1SymmetricCongurations:Aerodynamics Theaerodynamicsareevaluatedforthesymmetriccongurationsinwhichthesweepoftherightwingisequivalenttothesweepoftheleftwing.Inthiscase,theonlydegreesoffreedomaretheinboardandoutboardangleswhicharesharedbyeachwing.Theaerodynamicsarecomputedusingavortex-latticemethodthatisdesignedtoconsiderthinairfoils[ 30 ].Asetofrepresentativedataispresentedthatisparticularlyinformativewithrespecttothemaneuversanticipatedforthisclassofvehicle. ThevariationofliftwithrespecttoangleofattackisshowninFig. 6-11 forarangeofsweepcongurations.ThedatashowthattheaircraftobtainsitshighestCLalongaridgelinecorrelatingtoequalbutoppositesweepofinboardandoutboard.Conversely,thisderivativedecreasessignicantlyforcongurationsofinboardandoutboardbeingbothsweptbackorbothsweptforward.Assuch,theliftismoredependentonangleofattackbyutilizingtheadditionaldegreeprovidedbytheelbowtoopposethesweepoftheshoulder. Anotherlongitudinalparameter,Cm,isshowninFig. 6-12 forthesweepcongurations.Thisparameterisdirectlyindicativeofthestaticstability;consequently,thepositivevaluesindicatetheaircraftbecomesmorestaticallyunstableasthewingsaregraduallysweptforward.Thisinstabilityisdemonstrativeofacenterofgravitywhichliesaftofthe 128
PAGE 129
Figure6-11. VariationofLiftwithAngleofAttackforSymmetricSweep neutralpoint.Thecenterofgravityisrelativetotheplacementofmasseswithin,oron,theaircraft,andthereforeshiftsaccordingtoeachmorphingconguration. Figure6-12. VariationofPitchMomentwithAngleofAttackforSymmetricSweep Thedamping-in-rollderivative,towhichClpiscommonlyreferred,isshowninFig. 6-13 forthecongurationspace.Arollratecausesvariationsinangleofattackalongthespanofthewingwhichcreatesarollingmoment.Thisderivativeisnegativeforallsweepcongurations,withthelargestvaluedmagnitudesoccurringinregionscorrespondingtocongurationswithequalbutoppositesweepoftheinboardandoutboardsections.Themagnitudedecreasesforcongurationswithinboardandoutboardbeingbothforwardsweptorbothbackwardsweptwhichsuggestsapotentialitytoauto-rotateorspin. ThevehiclehasdirectionalstaticstabilityasevidencedbythedatainFig. 6-14 forallsymmetriccongurations.Thisdatarelatesthederivativeofyawmomentwithangle 129
PAGE 130
Figure6-13. VariationofRollMomentwithRollRateforSymmetricSweep ofsideslipwhosepositivitydemonstratesthestabilitycondition.Thestabilityisincreasedasthebackwardsweepincreasesbecausethestabilizingcontributionsofthefuselageandverticaltaildominateasthewingloseseectiveness. Figure6-14. VariationofYawMomentwithAngleofSideslipforSymmetricSweep 6.3.2SymmetricCongurations:FlightDynamics Linearizedmodelsoftheightdynamicsarecomputedbyrelatingtheaerodynamiccoecientstothestandardequationsofmotionforight[ 33 ],asgiveninFig. 6-15 .Theselinearizedmodelshavedecoupledstatesthatallowseparateanalysisoflongitudinaldynamicsandlateral-directionaldynamics.Modelsarecomputedforeverysymmetriccongurationintherangeofsweepanglestoindicatethevariedstabilityproperties. Thelongitudinaldynamicsarestable,asshowninFig. 6-16 ,forthemajorityofobtainablecongurations.Largevaluesofforwardsweepfortheinboardrequirealargevalueofbackwardsweepfortheoutboardtomaintainstability.Thesweepof 130
PAGE 131
Figure6-15. Modeltostate-spaceowchart theoutboardsectionisallowedtodecreaseastheinboarddecreasesitsforwardsweep.Eventually,thevehiclecanremainstabledespiteasmallvalueofforwardsweepfortheoutboardaslongastheinboardhasalargevalueofbackwardsweep. Figure6-16. NumberofUnstablePolesofLongitudinalDynamicsforSymmetricSweep Stabilityofthelateral-directionaldynamics,asshowninFig. 6-17 ,isachievedforasmallsetofcongurations.Theonlyregionofstabilitycorrespondstocongurationswithlargevaluesofbackwardsweepofbothinboardandoutboard.Theoneunstablepole,showninFig. 6-17 ,correspondstoaclassicallydenedspiralmodethatiscommonlyfoundtobeunstablewithalargetimeconstant. SomemodalpropertiesofthelongitudinaldynamicsarepresentedinFig. 6-18 toindicatethenumberofcomplexpoles.Eachpairofpolesrelatestoanoscillatorymodeso 131
PAGE 132
Figure6-17. NumberofUnstablePolesofLateral-DirectionalDynamicsforSymmetricSweep responsecharacteristicscanbedirectlyinferred.Inthiscase,thevehicledemonstratesaclassicalsetofphugoidandshort-periodmodesforthemajorityofcongurationsincludingallthosewithbackwardsweepoftheoutboardsections.Thephugoidmodeislostastheoutboardsectionsincreaseinforwardsweepuntileventuallyeventheshort-periodmodeislostforlargevaluesofforwardsweepfortheoutboard.Itcanbesaidthattheintroductionofunstablepoles,asshowninFig. 6-16 ,isdirectlyrelatedtothelossofbothoscillatorymodes,asshowninFig. 6-18 ,orviceversa. Figure6-18. NumberofOscillatoryPolesforLongitudinalDynamicswithSymmetricSweep ThenumberofoscillatorypolesisshowninFig. 6-19 forthelateral-directionaldynamics.Itisseenthatvehicleretainstwo-oscillatorypolesregardlessofthesweepconguration.Therefore,itcanbeinferredthatthevehiclehasaclassicdutchrollmode 132
PAGE 133
forallcongurations.Itcanbesaidthattheintroductionofunstablepoles,asshowninFig. 6-17 ,isnotcausedbyachangeinmodenature. Figure6-19. NumberofOscillatoryPolesforLateral-DirectionalDynamicswithSymmetricSweep 6.3.3AsymmetricCongurations Theaerodynamicsarealsocomputedforasetofasymmetriccongurationsinwhichtherightwingandleftwinghavedierentsweep.Theindependenceofinboardandoutboardoneachwingpresentsasetofcongurationswith4degreesoffreedom;consequently,thedatamustberestrictedtofacilitatepresentation.Theaerodynamicsarepresentedhereforcongurationsinwhichtherightwingisxedwith1=2=0andtheleftwingismorphedfrom-30degto30deginboththeinboardandoutboard. Asetofstandardparameterscanbecomputedtodirectlycomparetheaerodynamicsofsymmetricandasymmetriccongurations.Thevariationwithangleofattackforlift,showninFig. 6-20 ,andmoment,showninFig. 6-21 ,canbecomparedwithFig. 6-11 andFig. 6-12 ,respectively.Theclearsimilaritybetweenthesymmetricandasymmetricvaluesindicatessomerelationshipbetweenthecongurationscanbeinferred.Inparticular,thevariationcausedbysweepingbackasinglewingaresimilarinnaturetothevariationcausedbysweepingbackbothwings.Themagnitudeissmallerwhensweepingbackthesinglewingsosomelossofeciencyissuggested;however,thestabilityderivativesdisplaythesameshapeforeachsituation. 133
PAGE 134
Figure6-20. VariationofLiftwithAngleofAttackforAsymmetricSweep Figure6-21. VariationofPitchMomentwithAngleofAttackforAsymmetricSweep ThevariationofrollmomentwithrollratecanbecomparedinFig. 6-22 withFig. 6-13 alongwithvariationofyawmomentwithangleofsideslipshowninFig. 6-23 andFig. 6-14 forasymmetricandsymmetriccongurations.Again,thevariationsaresimilarinnatureforeachsetofcongurationssuggestingasimilarityinowphysicsbutalossofeciencyintheeect. Figure6-22. VariationofRollMomentwithRollRateforAsymmetricSweep 134
PAGE 135
Figure6-23. VariationofYawMomentwithAngleofSideslipforAsymmetricSweep Anadditionalsetofaerodynamicparametersarecomputedfortheasymmetriccongurationsthatarenullforthesymmetriccongurations.Theseparameters,whichareshowninFig. 6-24 ,representthecouplingbetweenlongitudinaldynamicsandlateral-directionaldynamics.Thedatashowsthatsweepingtheleftwingcausesadramaticincreaseinmagnitudeoftheseparameters.Sucharesultisexpectedsincetheseparametersreecttheasymmetrythatincreaseswithsweep. Figure6-24. VariationofCoupledAerodynamicsforAsymmetricSweep 6.3.3.1Flightdynamics Asetofmodelsthatrepresenttheightdynamicsarealsocomputedfortheasymmetriccongurations.Theselinearizedmodelsdonothavelongitudinalparametersdecoupledfromlateral-directionalparameters;consequently,theanalysismustconsiderasinglecoupledsystem. Thedynamicsareunstable,asshowninFig. 6-25 ,foranycongurationofasymmetricsweep.Thesystemisshowntohaveoneunstablepoleforthemajorityofcongurations 135
PAGE 136
andthreeunstablepolesforasmallregion.Thesmallregionisindicatedbyalargeforwardsweepoftheinboardsectionandaequaldisplacementofsweeparoundtheoutboardneutralposition. Figure6-25. NumberofUnstablePolesforDynamicswithAsymmetricSweep ThenumberofoscillatorymodesispresentedinFig. 6-26 todemonstratesomepropertiesofthevehiclemotion.Inthiscase,thevehiclehastheclassical3oscillatorymodesformostcongurationswhentheoutboardhasneutralorbackwardsweep.Astheoutboardissweptforwardoftheneutralposition,aregionofmodeswappingiscreated.Thisregionofforwardsweepfortheoutboardsectionindicatesthatamodeisgainedorlossstrictlydependingonthesweepoftheinboardsection. Figure6-26. NumberofOscillatoryPolesforDynamicswithAsymmetricSweep 6.3.3.2Modecharacterization Themodesofarepresentativecongurationischaracterizedtodemonstratethecoupledmotionwhichresultsfromanasymmmetricsweepofthewings.Theconguration 136
PAGE 137
ischosentohaveastraightwingontherightwithnosweepso1=2=0andastraightwingontheleftwithbackwardssweepso3=4=15deg.Theeigenvaluesgeneratedfromthisconguration,asshowninTable 6-3 ,indicatesevenstablepolesandoneunstablepolewhichisadistributionexpectedfromFig. 6-25 Table6-3. Setofeigenvalues Eigenvalues -17.59426.401i-37.202-2.78413.394i-0.1920.685i0.0610 FromTable 6-3 ,itcanbeseenthatthedynamicshavetwonon-oscillatorymodes.Thetimeconstantsofthesemodes,asshowninTable 6-4 ,indicatestheonemodehasastableconvergenceandtheotherhasanunstabledivergence.Thestableconvergenceisatleasttwoordersofmagnitudefasterthantheunstabledivergence. Table6-4. Timeconstantsofnon-oscillatorymodes ModeEigenvalueTimeConstant 1-37.200.026920.061-16.393 Theightmotionassociatedwitheachofthesemodesisdeterminedbythemodeshapes.SuchshapesaregiveninTable 6-5 todescribetherelativevalueofeachstateduringtheresponse.Theconvergenttermischaracterizedbymostlyrollratewithminorcontributionsfromangleofattack,rollangleandpitchrate.Thismodeissimilarinnaturetotheclassicallydenedrollmode.Thedivergenttermischaracterizedbyfullycoupledmotioninwhichtherollangleisvaryingalongwithprimarilytheyawrate,butalsotheforwardvelocity.Thismodeissimilarinnaturetotheclassicallydenedunstablespiralmode. 137
PAGE 138
Table6-5. Modeshapesofnon-oscillatorymodes statemode1mode2 forwardvelocity-0.0007-0.1324angleofattack-0.03980.0163pitchrate-0.05380.0003pitchangle-0.00140.0053angleofsideslip-0.00030.0100rollrate0.99710.0557yawrate-0.02310.3811rollangle-0.02680.9131 Table 6-3 alsoindicatesthattheightdynamicshavethreeoscillatorymodes.ThevaluesofnaturalfrequencyanddampingaregiveninTable 6-6 foreachmode.Themodewiththelowestnaturalfrequencyisunstablewhiletheothermodesarestable. Table6-6. Modalpropertiesofoscillatorymodes ModeEigenvalueFrequency(rad/s)Damping 3-17.59426.401i31.730.5464-2.78413.394i13.680.2035-0.1920.685i0.710.270 Theeigenvectorsassociatedwiththesemodesarealsocomplexsothemagnitudeandphaseofeachmodeshapeisusedtoanalyzetherelationshipbetweenstates.Theresultingdata,giveninTable 6-7 ,showsthatmodes3and5areprimarilydominatedbylongitudinalmotionwithonlyasmallcouplingtothelateral-directionalmotion,whilemode4isthedirectopposite.Suchmotionisnotentirelyunexpectedsinceeventhesymmetriccongurationshadoscillatorymodesaectingboththelongitudinalandlateral-directionaldynamics.Assuch,mode3haspropertieswithsomesimilaritytoashort-periodmode,whilemode4haspropertiessimilartoadutch-rollmodeandmode5similartoaphugoidmode. 6.3.3.3Crosswindrejection Sensorpointinginurbanenvironmentsisaprimemissionforwhichmicroairvehiclesarebeingdeveloped.Crosswinds,bothsteady-statewindandtime-varyinggusts,presentasignicantchallengetomaintainingsensorpointingduringight.Thecommonapproach 138
PAGE 139
Table6-7. Modeshapesofoscillatorymodes Mode3Mode4Mode5statemagnitudephase(deg)magnitudephase(deg)magnitudephase(deg) forwardvelocity0.02-42.60.0195.30.990.0angleofattack0.49-102.60.0536.10.08-179.1pitchrate0.620.00.0787.70.050.2pitchangle0.02-123.70.005-14.00.07-105.5angleofsideslip0.0002-61.20.0781.30.0002-20.3rollrate0.61-28.50.24-66.50.0051.5yawrate0.02-129.10.970.00.003-82.8rollangle0.02-152.20.02-168.30.007-104.2 tosensorpointingdespitecrosswindsisturningintothewindandcrabbingdownrangetoperiodicallypointthesensor;however,suchanapproachiscertainlynotoptimalduetothelackofcontinuouscoveragebythesensoralongthedesiredlineofsight. Asymmetricwing-sweepcanenhancetheabilitytoperformsensorpointinginthepresenceofsuchcrosswinds.Inparticular,onewingcanbesweptdownwindwhileonewingissweptupwind.Theaircrafthas,inasense,rotatedthewingsintothewindwhilethefuselageremainspointedinitsoriginaldirection.Thischangeintheeectivesideslipanglecaneasilybeillustrated,asseeninFig. 6-27 Figure6-27. EectiveAnglesofSideslip Theangleofsideslipatwhichtheaircraftcantrimisanindicatoroftheamountofcrosswindinwhichtheaircraftcanmaintainsensorpointing.Arepresentativedemonstration,showninFig. 6-28 ,presentsthemaximumpositivevaluesforangleof 139
PAGE 140
sideslipatwhichtheaircraftcantrim.Thewingsareconstrainedinthisdemonstrationsuchthatinboardandoutboardanglesareidenticalwhichlimitsthedegreesoffreedomandfacilitatespresentation.Also,eachconditioncorrespondstothelargestangleofsideslipatwhichtheaircraftcantrimgivendeectionlimitsof15degfortherudderandelevatoralongwithaileron. Figure6-28. MaximumAngleofSideslipatwhichAircraftcanTrim ThedatainFig. 6-28 demonstratesthatwingsweepisbenecialforsensorpointing.Specically,aforward)]TJ /F1 11.955 Tf 9.3 0 Td[(30degsweepoftheleftwingandabackward30degsweepoftherightwingallowsanangleofsideslipof44degtobemaintained.Thismaximumangledecreasesastheleftwingdecreasesitsforwardsweepandtherightwingdecreasesitsbackwardsweep.Thevehicleiseventuallyunabletotrimatanypositiveangleofsideslipwhenthebothwingsaresweptbackward. 6.4DynamicProperties 6.4.1MissionScenario Thevariablewing-sweepvehicleisdesignedforsurveillanceinurbanoperations.Inparticular,itisdesignedtoallowightinconstrainedareaswithlimitedairspace.Arepresentativemissionisplacingasensorintoawindowonabuildingwhichisclosetootherbuildings. 140
PAGE 141
6.4.1.1Divemaneuver Themissionproleforsuchamissioninvolvesseveralsegments.Theinitialightisstandardstraight-and-leveloperationatanaltitudeabovethebuildings.Uponreachingtheopeningbetweenthosebuildings,thevehicleentersasteepdivetorapidlydecreasealtitudewithoutincurringmuchforwardorsidetranslation.Theaircraftceasesthediveandreturnstostraight-and-levelightasquicklyaspossiblewhenthealtitudereachesthedesiredvalueassociatedwiththewindow. Themorphingvariestoincreaseperformanceofthemetricsassociatedwitheachmissionsegment.Theinitialightusesacruisecongurationwiththewingshavingnosweep.Thedivethenutilizesafullyswept-backconguration.Finally,thevehiclereturnstoanominalcongurationforenteringthewindow.Ineachcase,theaircraftusesasymmetricsweepofthewingstoavoidanycouplingassociatedwithasymmetriccongurations[ 46 ]. 6.4.1.2Turnmaneuver Themissionproleforsuchamissionagaininvolvesseveralsegments.Theinitialightisstandardstraight-and-leveloperationatanaltitudewithinthebuildings.Uponreachingadesiredopeningbetweenbuildings,thevehicleentersacoordinatedturntorapidlychangeheadingwithoutincurringmuchchangeinaltitude.Theaircraftceasestheturnandreturnstostraight-and-levelightasquicklyaspossiblewhentheheadingreachesthedesiredvalueassociatedwiththewindow. Themorphingagainvariestoincreaseperformanceofthemetricsassociatedwitheachmissionsegment.Theinitialightusesacruisecongurationwiththewingshavingnosweep.Theturnthenutilizesanequal-but-oppositecongurationarrangement.Anexampleofthisequal-but-oppositemorphingwouldbetheleftwingfullyforwardandtherightwingfullyswept.Finally,thevehiclereturnstoanominalcongurationforenteringthewindow.Itshouldbenotedthattheasymmetricmorphingcausesthedynamicstoexperiencecross-coupling. 141
PAGE 142
6.4.2MassDistribution Anelementalbreakdownoftheaircraft'smassdistributionallowsformoreaccurateinertialmomentsandratestobecomputed.Thesemassesarerepresentedaspointmassesandarelocatedatsomedistancefromtheaircraft'scenterofgravity.Thecenterofgravityisafunctionofwingmorph;therefore,ittranslatesalongthethree-dimensionalbody-axisaccordingly.Themagnitudeofthistranslationalchangeisfoundtobenegligiblethroughoutthedesiredmorphingrange,sinceverylittlemassisactivelymovedwhensweepingthewings,andthereforeassumedstationaryforthismodel. Adirectresultofhavingthisxedcenterofgravityisthattheaircraft'snon-dynamicpointmassesprovideforaconstantinertialmoment,whereonlythedynamicpointmasses,suchastheleftandrightwingsections,createanon-constantmoment.Theoverallmagnitudeofthisnon-constantmomentisdirectlydependentonhowtheaircraftisbeingmorphed.Iftheaircraft'swingsaresymmetricalwithrespecttothecommonlyassignedxz-plane,thenonlythexzinertialproducttermappears,whereas,ifthewingsareasymmetricalwithrespecttothexz-plane,thenallthreeinertialproducttermswillappear. Thewingisdividedintoseparateinboardandoutboardsections,whereacentroidlocationisfoundforeach.Basedonhowthewingismorphed,athreedimensionalaverageistakenbetweentheinboardandoutboardcentroids.Thisaverageistakenwithrespecttothecenterofgravityandisrepresentativeofanoverallcentroidlocationfromwhichthatwing'spointmassislocated. Asimplediagramillustratingtheaircraft'ssectionaldistributionandcenterofgravitylocationisshowninFigure 6-29 .NotethatinFigure 6-29 ,thecoordinatesystemisuniquetothemodelingprogramandisorientedoppositetotheearlierdenedaircraftbody-axis.ThisgureisaccompaniedbyTable 6-8 ,whichliststhemassesforeachsection.ThesemassesarethenusedtocalculatetheindividualinertialmomentslistedinTable 6-9 142
PAGE 143
Figure6-29. PointMassLocations Table6-8. Individualpointmasses MB(g)MF(g)MM(g)MTB(g)MVT(g)MHT(g)MRW(g)MLW(g) 1302955015884545 Table6-9. Characteristicsofelementsgivenascentroidposition(in)andmomentsofinertia(gin2) elementSymbolXYZIxxIyyIzzIxyIxzIyz batteryB-2.50.0-1.753981211812.50.0-5960.0fuselageF0.00.0-1.759039030.00.00.00.0motorM4.50.01.5113112510130.0-3380.0tailboomTB8.50.0-3.0135121910840.03830.0verticaltailVT13.50.00.00.0145814580.00.00.0horizontaltailHT13.50.0-3.072153014580.03240.0 6.4.3ManeuverAssumptions Asetofplantmodelsaregeneratedtorepresenttheightdynamicsateachconguration.Essentially,theforcesandmomentsaectingthevehiclearecomputedusinganassumptionofsteady-stateconditions.Theresultingmodelsdonotproperlyrelatetheunsteadyaerodynamicsbutwillincludethedominantsteady-stateaerodynamics.Thetime-varyinginertiasarethenintroducedtoaccountforthemorphingusingtheexpressionsderivedinSection 2.3.1.2 143
PAGE 144
6.4.4DiveManuever 6.4.4.1Modeling Thevehicleisconstrainedinthismaneuvertosimplifythecongurationspace.Thephysicalvehiclehas4degrees-of-freedominthateachinboardandoutboardcansweepindependentlyontheleftandrightsides;however,thissimulationconstrainstheleftandrightwingstosymmetricsweepwiththeoutboardhavingtwiceasmuchsweepastheinboard.Thisconstraintlimitsthesystemtoasingledegree-of-freedomwhichisappropriateforthelongitudinalnatureofthealtitudechangeassociatedwiththemission. Theightdynamicsforeachcongurationaretrimmedforstraightandlevelight.Actually,thethrustisheldconstantforeachofthesecongurationstonotethepropulsionisnotaectedbythemorphing.Suchanapproachisparticularlyusefulforrelatingmodelsthatmaybetrimmedatvariousairspeeds.Inthiscase,thethrustanddragwereheldconstantsothatthetrimroutinefoundthecorrectairspeed,asshowninFigure 6-30 ,torelateeachmodel. Figure6-30. ChangeinVelocityBasedonSymmetricMorphing:0deg( {2{ ),5deg( {/{ ),10deg( {{ ),15deg( {{ ),20deg( {.{ ),25deg( {{ ),30deg( {4{ ) Figure 6-30 relatestheoveralldragtovelocitywhereeachtrendline(designatedbylinetype),representsamorphingconguration.AconstantthrustoftwoNewtonsischosenand,thusalineisdrawntorepresentthisvalue.Itisshownthatasthewingsaremorphedfurtherback,theintersectionatwhichthetrendlinescrosstheconstant 144
PAGE 145
thrustlinesteadilyincreases.ThispointofintersectionrepresentsthevelocitynecessarytomaintaintwoNewtonsofdragatthatconguration.Overall,Figure 6-30 suggeststhatbysweepingthewingsback,highervelocitiescanbeattained,whilemaintainingaconstantdrag(equaltothrustintrimmedcases).Thistrendisasensibleresultduetothefactthatlesssurfaceareaisexposedtotheoncomingairowwhilethewingsaremorphedbackward. Thevelocitiesfoundateachintersectionarethenusedascorrespondingtrimvelocitieswhenperformingthedivemaneuver. 6.4.4.2Altitudecontroller Acontrollerisformulatedforthisaircrafttoenableamaneuverenvisionedinitsmissionprole;namely,acompensatorwillcommandthevehicletotrackadesiredaltitudeprole.Amulti-looparchitectureisusedtorepresentthecontrollerwhichwillbederivedinamodularapproach. Apairofcontrollersareactuallycomputedforthisaircraftsuchthatonecontrollerisappropriateforthenominalcongurationwhiletheotherisappropriatefortheswept-backconguration.Essentially,thecontrollersarederivedusinganassumptionofinstantaneousmorphing.Suchanapproachsimplyswitchesbetweenfeedbackgainstomatchtheassumedswitchbetweennominaldynamicsandswept-backdynamics.Obviouslythemorphingisnotinstantaneous;however,thisassumptionisnotoverlyunreasonablegiventhefastrateofmorphingontheaircraft. Aninner-loopcontrollerisderivedtotrackcommandstopitchrate.Thiscompensatoriscomputedusingalinear-quadraticregulator[ 87 ]usingafeedbackelement,K,andafeedforwardelement,k,alongwithanintegrator.Actually,thedesignisbasedonshort-perioddynamicstoavoidthepolesandzeroesassociatedwiththephugoiddynamics.Theguaranteeofnosteady-stateerrorinresponsetoastepcommandisonlyassociatedwiththeshort-periodmodelbutthefull-orderdynamicsusedinthesimulationstillshowanacceptableresponse. 145
PAGE 146
Anouter-loopcontrollerisderivedtoaecterrorinaltitude.Thedierencebetweencommandedaltitudeandmeasuredaltitudeisprocessedthrougharst-orderlter,F,toshapethephaseproperties[ 115 ].Theresultingsignalissomewhatindicativeofapitchcommandsoitsderivativeisusedasapitch-ratecommandfortheinner-loopcontroller. Theresultingclosed-loopsystem,asshowninFigure 6-31 ,incorporatesthevariouselementsusingappropriatefeedbackinthemulti-looparchitecture.Anadditionalfeedforwardelement,Z,isincludedinthedesigntoaectthepitchduringtheresponse. a F s 1 s k P K 6 6 ? Z ?huwq Figure6-31. Closed-LoopBlockDiagram 6.4.4.3Time-varyingdynamics Thesimulationofthemorphingaircraftmustproperlyaccountforthetime-varyingdynamics.Essentially,themorphingiscommandedtochangethroughoutthesimulationsotheightdynamicsmustchangeaccordingly.Theplant,P,inFigure 6-31 ,isthussomewhatcomplicatedinitsimplementation. Adiscretizedtypeofmorphingisutilizedinthesimulationtorepresentthesymmetricwing-sweepvariations.ThephysicalaircraftshowninFigure 6-9 hasasweepthatvariesasacontinuousfunctionoftime;however,adiscretizedversionofthismorphingsimpliesthesimulationwithoutlossofgenerality.Inthiscase,thedynamicsassumea5-degreesweepcanbeaccomplishedinstantaneouslybuteach5-degreeincrementmustbeseparatedbysomeminimumtime. 146
PAGE 147
Astandardstate-spaceformulationisusedtorepresenttheplantasshowninFigure 6-32 .Theplantdynamics,whichvarywithmorphingpositionandrate,aregivenbythequadrupleoffA;B;C;Dg. u U B 1 s -y X A 6 D ? Figure6-32. PlantModelwithTrimLogic TheelementofXisusedtoaccountforthechangeintrimconditionsforeachconguration.Considerthatatanypointintime,t,thetotalstatevalue,x(t),isactuallytheadditionofatrimvalue,xo(t),andperturbation,x(t),suchthatx(t)=xo(t)+x(t).Thistotalstatemustremaincontinuousdespitethemorphingeventhoughthetrimvalue,xo(t),willchange.Sincetheplantmodelusesstateperturbationsasafeedback,thenthelogicofXissuchthatx(t+t)=xo(t)+x(t))]TJ /F5 11.955 Tf 11.2 0 Td[(xo(t+t)forsomeintegrationstep-sizeoft. TheelementofUissimilarinnaturetoX.Inthiscase,theelementisusedtoreectthevariationsinelevatorpositionassociatedwithtrim. 6.4.4.4Simulation Thewingcongurationaltersduringthemissiontoreectthedesiredperformance.Therstcongurationhasnosweeptoreectacruiseprole.Thewingsarethensweptbacktothemaximumvaluewhenthediveisinitiated.Uponreachingthedesiredaltitude,thevalueofthesweepisreducedtoreturntoaproleforenteringthewindow.Theangleofthesweep,asshowninFigure 6-33 ,isvariedforsimulationsbasedonfastmorphingorslowmorphing. 147
PAGE 148
Figure6-33. MorphingCongurationforFastMorphing( | ),SlowMorphing( )-222()-222()]TJ ET 0 G 0 g BT /F1 11.955 Tf 438.74 -186.8 Td[() Thealtitudevariation,showninFigure 6-34 ,isthefundamentalmetricusedtoevaluateperformance.Inthiscase,themorphingvehiclesareabletochangetherequiredaltitudeandreturntostraight-and-levelightwithin2.2s.Thefastandslowmorphingaresimilarfortherstsecondbutthenbegintodivergesomewhat.Thefastmorphinghasreachedafullyswept-backcongurationatthistimesoitsresponseahassomewhatlargerovershootthanthatseenwiththeslowmorphing.Thisobservationisadirectresultofthefastmorphingcaseacquiringahighervelocitysoonerandretainingitforalongerperiodoftime.Twoothercasesareconsiderwheremorphingisnotutilized.Thesecasesincludedivemaneuversperformedwiththenominalstraight-wingandfullysweptcongurations.ItisseeninFig. 6-34 thatthenominalcongurationreachesthethedesiredaltitudetheslowestbuthasthesmallestovershoot,whilethefullysweptcongurationreachesthedesiredaltitudemuchfaster,yethasthelargestovershoot. ThestatesassociatedwithpitchareshowninFigure 6-35 inresponsetothealtitudecommandwhilemorphing.Theslowmorphing,incomparisontothefastmorphing,incursagreaterpitchangleduringtheinitialresponsebutthenincursasmallerpitchangleasthevehiclereachesitsnalaltitude.ThisbehaviorcorrelateswiththealtituderesponseofFigure 6-34 148
PAGE 149
Figure6-34. AltitudeinResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( )-222()-222()]TJ ET 0 G 0 g BT /F1 11.955 Tf 139.37 -170.84 Td[(),FixedStraight( )-221(\001 ) Figure6-35. PitchAngle(left)andPitchRate(right)inResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( )-222()-222()]TJ ET 0 G 0 g BT /F1 11.955 Tf 279.35 -357.97 Td[(),FixedStraight( )-222(\001 ) Thelateral-directionalstatesarefoundnottovarywithalongitudinaldivemaneuver.Thisresultissomewhatexpected,duetothefactthattheaircraftissymmetricallymorphingandthusthecross-couplinginertiasarecancelledoutduetosymmetry. Finally,theelevatorangleisshowninFigure 6-36 todemonstratetheresponsedoesnotincurexcessiveactuation.Themorphingcertainlyinuencestheelevatorinthatthecontroleectivenessdecreasesasthewingsaresweptback.Assuch,theslowmorphinghastheslowestspeedbutalsousesthesmallestrotationofelevator. 6.4.4.5Missionevaluation Theclosed-loopsystemisnotabletosuccessfullycompletethemissionusingonlythenominalcontrollerdesignedforthenominalwingconguration.Thevehicleisabletodivebetweenthebuildingsandsuccessfullychangealtitudewithininanitetimeasshownin 149
PAGE 150
Figure6-36. AltitudeinResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( )-222()-222()]TJ ET 0 G 0 g BT /F1 11.955 Tf 139.37 -165.43 Td[(),FixedStraight( )-221(\001 ) Figure 6-34 ;however,ifconstrainedtoacertainspaceandtime,theresultingightpathsassociatedwithmorphingmissthewindowandintersectthesideofthebuilding,asseeninFigure 6-37 .Thisfailuredirectlyresultsfromtheinabilitytoaccountfortime-varyingeectsinthecontrollerdesign.NotethatinFigure 6-37 thecoordinatesarebasedonanEarth-xedframewithaxesdenedbyX,Y,andH. 6.4.4.6Eectsoftime-varyinginertia Theightcontrollerneedstobealteredtocompensateforthetime-varyingparametersassociatedwithmorphing.Althoughmorphingintroducesinertialratesintothedynamics;theseratesarefoundtohavelittleeectontheoverallplantdynamicsandthereforealtertheightpathslightly,asseeninFig. 6-38 .Itshouldbenotedthattheinertialrates,asshowninEquationrefeq41,arecalculatedfromwingscontaininglessthanfteenpercentoftheaircraft'soverallmass.Whenthemassesofthewingsegmentsareincreased(orthetimetomorphisdecreased,asshownbythefastmorphinginFig. 6-38 ),thenthecontributionsfromtheinertialratesbecomemoreprominentintheoveralldynamics.Theresultingchangeindynamicswouldaltertheightpathfurther,thusillustratingtheimportanceofinertialrates. Dierentapproachesmaybetakentoaltertheightcontroller,buteachhasit'sownchallenges.Acompensatortoregulatethepitchisdicultbecausethegainsmustbeoptimizedtothedynamicsbutthosedynamicsarerapidlychangingduringthemorphing. 150
PAGE 151
Figure6-37. AltitudeinResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( )-222()-222()]TJ ET 0 G 0 g BT /F1 11.955 Tf 139.37 -468.51 Td[(),FixedStraight( )-221(\001 ) Alternatively,atrackingcontrollercanbeusedafterthemaneuverwhenthedynamicsareknownandxedbutthevehiclemustre-hometowardstheoriginalwaypointorre-locatethewindowusingvisionfeedbackandthencomputeanewtrajectory. 151
PAGE 152
Figure6-38. AltitudeinResponsetoFastMorphing( | ),SlowMorphing( ::: ),FastMorphingWithoutInertia( )-221(\001 ),SlowMorphingWithoutInertia( )-222()-222()]TJ ET 0 G 0 g BT /F1 11.955 Tf 456.7 -396.51 Td[() 6.4.5CoordinatedTurnManeuver 6.4.5.1Modeling Thevehicleisconstrainedinthismanuevertoagainsimplifythecongurationspace.Thismaneuverconstrainstheleftandrightwingstoequal-but-oppositeasymmetricsweepwiththeoutboardhavingnorelativesweepcomparedtotheinboard.Thisconstraint,likethedivemaneuver,limitsthesystemtoasingledegree-of-freedomwhichisappropriateforthelateral-directionalnatureofthechangeinturnperformanceassociatedwiththemission. 152
PAGE 153
Theightdynamicsforeachcongurationaretrimmedforsteadybankedight,andagainthethrustisheldconstantforeachofthesecongurations.Therefore,thetrimroutinefoundthecorrectairspeed,asshowninFigure 6-39 ,torelateeachmodel. Figure6-39. ChangeinVelocityBasedonAsymmetricMorphing:0deg( {2{ ),5deg( {/{ ),10deg( {{ ),15deg( {{ ),20deg( {.{ ),25deg( {{ ),30deg( {4{ ) Fig. 6-39 relatestheoveralldragtovelocitywhereeachtrendlinerepresentsamorphingconguration.AconstantthrustoftwoNewtonsisagainchosenasthedesiredvalue.Itisshownthatasthewingsareasymmetricallymorphed(equalbutopposite,wheretheangleismeasuredwithrespecttotherightwing),theintersectionatwhichthetrendlinescrosstheconstantthrustlinesteadilyincreases.ThispointofintersectionrepresentsthevelocitynecessarytomaintaintwoNewtonsofdragatthatconguration.Overall,Fig. 6-39 suggeststhatbyasymmetricallysweepingthewingsinanequalbutoppositemanner,slowervelocities(inabankedturn)areattainedwhilemaintainingaconstantdrag(equaltothrustintrimmedcases). Thevelocitiesfoundateachintersectionarethenusedascorrespondingtrimvelocitieswhenperformingthecoordinatedturnmaneuver. 6.4.5.2Turncontroller Thecoordinatedturnmaneuverusesabasicclosed-loopapproachforcontrollingthesystem,althoughanopen-loopdesigncouldbeused.Anopen-loopcontrollerwould 153
PAGE 154
besucientbecauseeachplantmodelispreviouslytrimmedforabankedturn.Themodelingtool,AVL,incorporatesaninternalcommandusedfortrimmingmodelsatauserdenedbankangle.Therefore,ifnopilotcommandsaregivenandallexternaldisturbancesareneglected,theaircraftwillremaintrimmedandcompletethemaneuver. Itisnotedthatthemodelsusedtosimulatethismaneuverareperturbationmodelsandthereforeproducestateperturbations.Asaresult,abasicfeedbackloopisincorporatedtomeasurethedierencebetweenthebankangleperturbationandzero.Azerocommandisgiventorepresentthedesiredperturbationfromtrim.Themeasureddierenceisthenmultipledbyaproportionalgainandfedintotheplantasanaileroncontrolcommand.Thisfeedbackloopisincorporatedtoguaranteecontinuitythroughoutmorphing,asdescribedfurtherinSection 6.4.5.3 .Theresultingclosed-loopsystemcanbeseeninFigure 6-40 u k P ?uwqpr Figure6-40. Open-LoopBlockDiagram 6.4.5.3Time-varyingdynamics Likeinthedivemaneuver,themorphingisagaincommandedtochangethroughoutthesimulationandthereforetheightdynamicschangeaccordingly.Althoughtheplant,P,inFigure 6-40 ,isslightlydierentfromtheplantmodelderivedforthedivemaneuverinFigure 6-31 ,itisstillsomewhatcomplicatedinitsimplementation. 154
PAGE 155
Adiscretizedtypeofmorphingisagainutilizedinthesimulationtorepresenttheasymmetricwing-sweepvariations.Theturnsimulationisidenticaltothedivesimulationinthatthephysicalaircrafthasasweepthatvariesasacontinuousfunctionoftimeandadiscretizedversionofthismorphingisusedinthesimulation. Astandardstate-spaceformulationisusedtorepresenttheplantasshowninFigure 6-41 .Theplantdynamics,whichvarywithmorphingpositionandrate,aregivenbythequadrupleoffA;B;C;Dg. u E A R B 1 s -y X A 6 D ? Figure6-41. PlantModelwithTrimLogic TheelementofXisagainusedtoaccountforthechangeintrimconditionsforeachconguration.Thesameconsideration(asthatforthedivesimulation)ismadetoaccountforthecontinuityofthetotalstatedespitethemorphing,aswellasthefeedbackofperturbations. TheelementsA,EandRaresimilarinnaturetoU,asseeninFigure 6-32 .Inthiscase,theelementsareusedtoreectthevariationsinaileron,elevator,andrudderposition,respectively,associatedwithtrim. 6.4.5.4Simulation Amissionischosenfortheturnmaneuver,similartothedivemaneuver,suchthatitwillagaintryytointoaonemeterwindow.Thewingcongurationaltersduringthemissiontoreectthedesiredperformance.Therstcongurationhasnosweepto 155
PAGE 156
reectacruiseprole.Thewingsarethenasymmetricallyswept(equalbutopposite)toachieveamaximummorphedcongurationconsistingoftheleftwinghavingafullforwardsweepwhiletherightwingisequalbutopposite.Themaximummorphedcongurationismaintaineduntiltheendofthemaneuverwhenthevalueofthesweepisreducedtoreturntoaproleforenteringthewindow.Theangleofthesweep,asshowninFigure 6-42 ,isvariedforeachsimulationbasedonfastmorphingorslowmorphing. Figure6-42. MorphingCongurationforFastMorphing( | ),SlowMorphing( )-222()-222()]TJ ET 0 G 0 g BT /F1 11.955 Tf 438.74 -334.56 Td[() Theturnvariation,showninFigure 6-43 ,isthefundamentalmetricusedtoevaluateperformance.Inthiscase,themorphingvehiclesareabletochangetherequiredturningradiusandreturntostraight-and-levelightata270degreeheadingchange.Thefastandslowmorphingaresimilarduringtherstandlastsectionsofthemaneuverbutvarysomewhatthroughouttheactualturn.Theresponseoffastmorphingmorecloselyresemblesthatofthenon-morpedsweptresponseduetothefactthefastmorphingreachesthefullysweptcongurationatanearliertime.Therefore,thefastmorphingresponsecontainsalargerportionproducedfromthesweptwingcongurationthanthatoftheslowmorphingresponsewhichtakeslongertoachievethesameconguration. ThelateralperturbationstatesareshowninFigure 6-44 astherollangleandrollrate.Itisseenthatthefastermorphinginboththerollangleandrateperturbations 156
PAGE 157
A B C Figure6-43. TurninResponsetoFastMorphing( | ),SlowMorphing( ::: ),FixedSwept( )-222()-222()]TJ ET 0 G 0 g BT /F1 11.955 Tf 139.37 -355.98 Td[(),FixedStraight( )-221(\001 ):A)ThecompleteturnproleB)Theturnproleat270degreesC)Theturnproleat180degrees achievehighermagnitudes.Therollangledisplaysthistrendthroughouttheentiremaneuver,whiletherollrateonlyhaslargermagnitudesatthetransientregions. Figure6-44. RollAngle(left)andRollRate(right)inResponsetoFastMorphing(|),SlowMorphing()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() 157
PAGE 158
ThedirectionalperturbationstatesinFigure 6-45 indicatetheyawrateandsideslipangle.Itisnoticedthatthefastermorphingincurslargerperturbationsinbothdirectionalstatesthroughouttheentiremaneuver. Figure6-45. YawAngle(left)andYawRate(right)inResponsetoFastMorphing(|),SlowMorphing()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() Itisnotedthattheturnmaneuverrequirestheaircrafttomorphasymmetrically,andthereforeresultsincoupleddynamics.ThestateperturbationsassociatedwithpitchareshowninFigure 6-46 inresponsetothesimplecontrolsurfacecommandwhilemorphing.Thefastmorphing,incomparisontotheslowmorphing,incursagreaterpitchrateperturbationatbothtransientregions,yetbehavesverysimilarduringthesteadystateregion.Incontrast,thepitchangleperturbationassociatedwithfastmorphinghasahighermagnitudethroughouttheentiremaneuver. Figure6-46. PitchAngle(left)andPitchRate(right)inResponsetoFastMorphing(|),SlowMorphing()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() 158
PAGE 159
6.4.5.5Missionevaluation Theclosed-loopsystemisnotabletosuccessfullycompletethemission.Thevehicleisabletoturnbetweenthebuildingsandsuccessfullychangeheadingwithinthedesiredtimeandairspace.However,whentheeectsofmorphingareintroduced,theresultingightpathmissesthewindowandintersectsthesideofthebuilding,asseeninFigure 6-47 .Thisfailuredirectlyresultsagainfromtheinabilitytoaccountfortime-varyingeectsinthecontrollerdesign.NotethatinFigure 6-37 ,thecoordinatesarebasedonanEarth-xedframewithaxesdenedbyX,Y,andH. Figure6-47. TurninResponsetoFastMorphing(|),SlowMorphing(:::),FixedSwept()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[(),FixedStraight()-221(\001) 159
PAGE 160
TheeectsofmorphingonthetrajectoryoftheightpathcanbetterbeseeninFigs. 6.4.5.4 6.4.5.4 6.4.5.6Eectsoftime-varyinginertia Itwasseeninthepreviousmissionthatinertialrateswereintroducedasaresultofmorphing.Duetothesymmetryoftheaircraftinthatmaneuver,mostinertialmomentsandthereforerates,werecancelledout.Inthecaseoftheturnmaneuver,alltheinertialmomentsandratesarekept.Thepresenceofthesetermshasamorepronouncedeectontheturnradius,asseeninFigure 6-48 ,thanthatseenfordiveinthepreviousmission,asseeninFigure 6-38 Figure6-48. TurninResponsetoFastMorphing(|),SlowMorphing(:::),FastMorphingWithoutInertia()-221(\001),SlowMorphingWithoutInertia()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() 160
PAGE 161
Theightcontrollerforthismissionalsoneedstobealteredtocompensateforthetime-varyingparametersassociatedwithmorphing.Simplyintroducingacompensatortoregulatethelateral-directionalstatesischallengingbecausethegainsmustbeoptimizedtothedynamicsbutthosedynamicsarerapidlychangingduringthemorphing.Itispointedoutthatthedynamicsarenowcoupled(duetoasymmetricmorphing)andthereforethecontrollermustalsocompensateforthelongitudinalstates.Atrackingcontrollermayalsobeusedbutthesamedicultiesarisethatwerepointedoutwiththedivemaneuver. 6.5Time-VaryingModalAnalysis 6.5.1Kamen'sMethod 6.5.1.1Statespacemodel Theintroductionofsymmetricmorphingvariestheaerodynamicsandassociatedvaluesofstabilityderivatives.Thesevariationsresultingfrommorphingthesweepsymmetricallyareprimarilyrestrictedtosomederivativesofverticalforceandallthederivativesofpitchmoment.Assuch,thestandardstate-spacerepresentationofthelinearizeddynamicsismodiedinEquation 6{1 toreectthedependencyofthesederivativesonthemorphinggivenas.Ithasbeenshown[ 20 ]thatastheorderofthedynamicsystemincreases,sodoesthediucltytorepresentthem(usingKamen'smethod).Therefore,themodalanalysisutilizingKamen'smethodwillbelimitedtothefourthordersystemdescribedbyEquation 6{1 .Thestatesofthissystemaregivenasuforforwardvelocity,wforverticalvelocity,qforpitchrate,andforpitchanglewhilethestabilityderivativesincludeXiforderivativeoflongitudinalforcewithrespecttotheistate,Ziforderivativeofverticalforcewithrespecttotheistate,andMiforderivativeofpitchmomentwithrespecttotheistate. 161
PAGE 162
266666664_u_w_q_377777775=266666664XuXwXq)]TJ /F5 11.955 Tf 9.3 0 Td[(gcos0ZuZw()u0)]TJ /F5 11.955 Tf 9.3 0 Td[(gsin0Mu()Mw()Mq()00010377777775266666664uwq377777775(6{1) Theanalysisoftime-varyingdynamicsusingthepolesdenedinEquation 3{14 requiresthestate-spacesysteminEquation 6{1 tobeformulatedasadierentialequation.Inthiscase,theequationinvolvingtheverticalvelocity,w,isconsidered.Theequationsinvolvingtheotherstatescanalsobederived;however,thepolesassociatedwiththeverticalvelocityaresucienttocharacterizetheentiresystemsinceanypolesetofobtainedbyatransformationofthesepoles. Thegeneralformofthefourth-orderexpressionforverticalvelocityisgiveninEquation 6{2 byincludingtime-varyingcoecients,A1;A2;A32R,whicharelengthyandthusdenedintheAppendix. d4w dt4+A3(t)d3w dt3+A2(t)d2w dt2+A1(t)dw dt+A0(t)w=0(6{2) ThecoecientsinEquation 6{2 havesignicantlydierentdependenciesonthemorphinginthateachdependsonthestabilitiesderivativesofZw;Mw;Mq;muwhichvarywithmorphingbutnoteachdependsontheratesofchangesofthesestabilityderivatives.ThevalueofA0dependsonboththerstderivativeandsecondderivativewithrespecttotimeforall4ofthesestabilityderivatives;conversely,A3dependsonlyontherstderivativeandsecondderivativewithrespecttotimeforMqandMu.TheeectofmorphingonA3canevenbefurthermitigatedbynotingthatA3dependsonthemorphingvaluebutnotthemorphingrateif_Mu _Mq=Mu Mq=)]TJ /F11 7.97 Tf 10.5 4.71 Td[(Zu uo.Assuch,thetimevariationsofcertainstabilityderivativesonlyaectcertaincoecientsdependingontherateofchangeofthemorphingtrajectory. 162
PAGE 163
Also,thedynamicshaveapotentialtoexperienceabifurcationthatreducestheorderofthedynamics.ThecoecientsinEquation 6{2 eachcontainafractionwiththesamedenominatorwhichiszeroforcertainvaluesofmorphing.ThelossoforderisequivalentlyviewedbymultiplyingEquation 6{2 bythisdenominatorsothatthedependenceond4w dt4isscaledbyzero. 6.5.1.2Morphingtrajectory TheightdynamicsofthevehicleshowninFigure 6-9 areanalyzedduringsymmetricmorphingfromabackwardsweeptohavingnosweep.Specically,thesweepvariedfrom+30degto0degin1secandthenremainsatasweepangleof0degforeachwing.Thismorphingwouldbevaluablewhentransitioningfromadivetostraight-and-levelightsimilarinamannertobiologicalsystemslikegullsandhawks.Thistransition,especiallywhenoperatingimmersedindenseobstaclessuchasurbanenvironments,maystillrequirerapidmaneuveringforpositioningalongwithgustrejectionsotheightdynamicsduringthemorphinremainofcriticalimportance. TheresponseofthestatesareshowninFigure 6-49 asaresultofthemorphing.Theaircraftisalineartime-varyingsystemsfortheinitial1sec;however,theresponsestillresemblesthetraditionalmodesforalineartime-invariantsystem.Thepitchrateandverticalvelocityshowahigh-frequencyresponsethatisheavilydampedtoresembletheshort-periodmode;conversely,theairspeedandpitchanglearedominatedbyalow-frequencyresponsethatislightlydampedtoresembleaphugoidmode. 6.5.1.3Time-varyingpoles Thetime-varyingpolesassociatedwithFigure 6-49 arecomputedtosatisfyEquation 3{14 andshowninFigure 6-50 alongwiththetime-invariantpolesthatignorethetime-varyingeectsofmorphing.Theseresultindicateseveralcharacteristicsoftime-varyingpoles.Considerthatthetime-invariantpolesassociatedwiththeshort-periodmoderemainsimilarforanyvalueofmorphing;however,thetime-varyingpolesofp41andp42,whichareinitiallyclosestinvaluetotheshort-periodpoles,decaytozeroatarate 163
PAGE 164
Figure6-49. LongitudinalStatesduringMorphingfrom+30degto0degover1sec:ForwardVelocity(upperleft),VerticalVelocity(upperright),PitchRate(lowerleft),PitchAngle(lowerright) similartodecayofthestateresponseduetodamping.Also,notethatthepolesofp43andp44showthelowmagnitudeandslowdecayassociatedwithaphugiodmode;however,thesepolesvaryduringtheinitialresponsewhichisdominatedbytheashort-periodresponse.Assuch,thetime-varyingpolesdierinnaturefromtime-invariantpolesinthattheshort-periodmodeandphugoidaresomewhat,butnotcompletely,distinctandthemagnitudeofthepolesdecaysastheresponsedecays. ThemodesdenedinEquation 3{9 associatedwitheachpoleinFigure 6-50 areshowninFigure 6-51 .Thedecompositionoftheresponseactuallydependsontheeigenvectorsandthesemodes,asopposedtothetime-varyingpoles,sotheymustbeconsideredwhenevaluatingtheightdynamics.Theindistinctseparationbetweentheshort-periodmodeandthephugiodmodethatisevidentinthepolesisalsoevidentinthemodes.Themodesof41and42showtheinitialvariationthatwouldindicateashort-periodresponse;however,thesemodesremainsignicantafterthedecaydueto 164
PAGE 165
Figure6-50. LinearTime-VaryingPoles(|)andLinearTime-InvariantPoles()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()duringMorphingfrom+30degto0degover1sec:RealPart(upperleft)andImaginaryPart(lowerleft)ofp41andp42,RealPart(upperright)andImaginaryPart(lowerright)ofp43andp44 short-perioddampingandthenshowvariationmoreconsistentwiththephugoidmode.Themodesfor43and44showsomeinitialinconsistencyaround1secbutthenarequiteconsistentwiththephugiodmode.Indeed,therealpartsof43and44areremarkablysimilartotheresponseofforwardvelocity,whichispredominatelyduetothephuoidmode,inFigure 6-49 Thenatureofthemodesagreeswiththemathematicalpropertiesthatrelatethemtoboththeresponseandthepoles.TheresponsesofFigure 6-49 areoscillatoryandindeedthepolesofFigure 6-50 arecomplexconjugatessothemodesofFigure 6-51 arealsocomplexconjugates.TherealandimaginarypartsofthemodesarenotedinEquation 4{18 tobe90ooutofphaseandindeedthisphasedierenceisseenforthealltimesofthephugoidresponseof43andtheinitialtimesofshort-periodresponseof41until0:2secwhenthedampinghascausedtheresponsetodecay.Also,thestateresponseshouldbeproportionaltotherealpartofthemodeasnotedinEquation 4{19 whichis 165
PAGE 166
Figure6-51. ModesAssociatedwithTime-VaryingPolesduringMorphingfrom+30degto0degover1sec:RealPart(upperleft)andImaginaryPart(lowerleft)of41and42,RealPart(upperright)andImaginaryPart(lowerright)of43and44 demonstratedbytheverticalvelocityinFigure 6-49 matchingtherealpartof1andtheforwardvelocityinFigure 6-49 matchingtherealpartof3. TheissueofstabilityisdirectlyindicatedbythemodesofFigure 6-51 .Thesemodesdemonstratethatthesystemduringthismorphingtrajectoryhasasymptoticstabilitysincethemagnitudeofeachmodedecaystozeroastimeincreases.ThisresultcorrelateswiththeresponsesshowninFigure 6-49 thatobviouslyreturntoequilibrium.Notethatoneguaranteeforasymptoticstabilityishavingnegativereal-partforthetime-varyingpole.TherealpartofthepolesinFigure 6-50 predominatelyassociatedwiththephugoidmodeareindeedalwaysnegative;however,therealpartofthepolespredominatelyassociatedwiththeshort-periodmodearesometimespositivesothemodemustbecomputedtoascertainstability. Finally,theeigenvectorsassociatedwitheachmodeofFigure 6-51 aregraphedinFigure 6-52 toshowtherelativeresponseofeachvehiclestateasnormalizedbythe 166
PAGE 167
verticalvelocity.Theseeigenvectors,similarlyasthethepolesandmodes,showbothshort-periodcharacteristicsandphugiodcharacteristicsbuteachisclearlydominatedbyonetypeofdynamic.Theeigenvectorofv1initiallyshowsshort-periodmotion,withlittlevariationinforwardvelocityandaphasedierenceof90obetweenpitchrateandverticalvelocity,untilthedampingdecaysthatmotionandthephugoidresponseisevident.Theeigenvectorofv2steadilytransitionstothephugiodresponsewhichisprimarilymotioninforwardvelocityandpitchanglewhichare90ooutofphase.Also,notethattheseeigenvectorsnearlyconvergetosimilarmagnitudesandphasesexceptfora90odierenceinphaseofthepitchanglebetweenv1andv2. Figure6-52. NormalizedEigenvectorsAssociatedwithTime-VaryingModesduringMorphingfrom+30degto0degover1sec:Magnitude(upperleft)andPhase(lowerleft)ofv1andMagnitude(upperright)andPhase(lowerright)ofv2 6.5.1.4Modalinterpretation AmodalinterpretationofthepolesinFigure 6-50 isconductedtorelatethesemathematicalconstructstostandardparametersassociatedwithightdynamics.Theparametersaredirectlycomputedfromthetime-invariantpoleswhiletheircounterparts 167
PAGE 168
fromthetime-varyingpolesresultfrominterpretationsthatrelatecharacteristicsoftheresponses. ThenaturalfrequenciesasshowninFigure 6-53 havesomecommonalitiesbutalsosomecleardierenceswhencomparingthetime-varyingpolesandthetime-invariantpoles.Thevaluesarereasonablyclosefortheentiretrajectorywhenconsideringthepolesassociatedwiththephugoidmode;however,thevaluesareonlycloseforashorttimewhenconsideringtheshort-periodmode.Thedierenceinnaturalfrequenciesfortheshort-periodmoderesultsfromtherelationshipofthetime-varyingpolestothestates.Essentially,theshort-periodpoleisinitiallyrelatingtheoscillatorybehavioroftheresponsebutthesignicantdecreaseinresponsemagnitudeduetodampingisactuallyreectedbythetime-varyingpoledecayingtozero. Figure6-53. NaturalFrequencyAssociatedwithLinearTime-VaryingPoles(|)andLinearTime-InvariantPoles()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()duringMorphingfrom+30degto0degover1sec:Poles1and2(left)andPoles3and4(right) TheenvelopethatboundstheresponsesareshowninFigure 6-54 .Theparametersaresimilarforthephugoidmodebutagain,aswiththenaturalfrequency,theparametersdierforthetime-varyingpolesandthetime-invariantpolesfortheshort-periodmode.Inthiscase,theenvelopeboundstheresponseofthepitchrateandverticalvelocitywhichdominatetheshort-periodresponsebutthen,afterthatmodehasdampedout,theenvelopereecttheboundonthepitchangle. 168
PAGE 169
Figure6-54. EnvelopeAssociatedwithLinearTime-VaryingPoles(|)andLinearTime-InvariantPoles()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()duringMorphingfrom+30degto0degover1sec:Poles1and2(left)andPoles3and4(right) ThedampingratioisshowninFigure 6-55 forthetime-varyingpolesandtime-invariantpoles.Thesevaluesaresmallforbothtypesofpolesandthusarereasonablyclose. Figure6-55. DampingRatioAssociatedwithLinearTime-VaryingPoles(|)andLinearTime-InvariantPoles()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()duringMorphingfrom+30degto0degover1sec:Poles1and2(left)andPoles3and4(right) 6.5.2O'Brien'sMethod 6.5.2.1Statespacemodel AsopposedtoKamen'smethod,itwasshowninSection 3.5 ,thatO'Brien'smethodcaneasilybeimplementedonhigherordersystems.Therefore,individual,eightstate,quasi-staticplantmodelsaretakenatperiodiccongurationsovertheentiremorphing 169
PAGE 170
range.Thesemodelsareshowntobe8x8realstatematrices.Eachperiodiccongurationrepresentesadiscretepointintime.Plantmatricesarethenbrokendownbyelements,whereeachelementisplacedintocorrespondingvectors.Theelementvectorsaretheneachanalyticallyttoauniquepolynomial.Byplacingthesefunctionsbackintotheiroriginalpositionswithintheplantmatrix,anequivalenttime-varyingmatrix(ormodel)canbeformed,asshowninEquation 6{3 A(t)=2666666666666666666664f11(t)f12(t)f13(t)f14(t)f21(t)f22(t)f23(t)f24(t)f31(t)f32(t)f33(t)f34(t)f41(t)f42(t)f43(t)f44(t)00f55(t)f56(t)f57(t)f58(t)f65(t)f66(t)f67(t)f68(t)f75(t)f76(t)f77(t)f78(t)f85(t)f86(t)f87(t)f88(t)3777777777777777777775(6{3) ThefunctionsdenedinEquation 6{3 arefoundtohaveapolynomialstructure,asshowninEquation 6{4 170
PAGE 171
f11(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(0:09t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:33f55(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(0:03t2+0:07t)]TJ /F1 11.955 Tf 11.95 0 Td[(0:77f12(t)=0:13t2)]TJ /F1 11.955 Tf 11.96 0 Td[(0:67t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:29f56(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(0:0013t2)]TJ /F1 11.955 Tf 11.95 0 Td[(0:002t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:0017f13(t)=)]TJ /F1 11.955 Tf 9.29 0 Td[(0:03t2)]TJ /F1 11.955 Tf 11.96 0 Td[(0:08t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:15f57(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(0:0006t)]TJ /F1 11.955 Tf 11.95 0 Td[(0:99f14(t)=0:03t)]TJ /F1 11.955 Tf 11.96 0 Td[(9:79f58(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(0:001t+0:41f21(t)=0:16t2)]TJ /F1 11.955 Tf 11.96 0 Td[(0:66t)]TJ /F1 11.955 Tf 11.96 0 Td[(1:49f65(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(31:3t2+643t)]TJ /F1 11.955 Tf 11.95 0 Td[(1013f22(t)=0:79t2)]TJ /F1 11.955 Tf 11.96 0 Td[(3:45t)]TJ /F1 11.955 Tf 11.96 0 Td[(8:95f66(t)=5:94t2)]TJ /F1 11.955 Tf 11.96 0 Td[(20:73t)]TJ /F1 11.955 Tf 11.95 0 Td[(25:23f23(t)=0:19t2)]TJ /F1 11.955 Tf 11.96 0 Td[(0:23t+22:41f67(t)=0:6t3)]TJ /F1 11.955 Tf 11.96 0 Td[(1:17t2)]TJ /F1 11.955 Tf 11.96 0 Td[(3:92t+14:76f24(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(0:12t2+0:51t+0:62f68(t)=0f31(t)=)]TJ /F1 11.955 Tf 9.29 0 Td[(0:63t3+5:96t2)]TJ /F1 11.955 Tf 11.95 0 Td[(5:12t)]TJ /F1 11.955 Tf 11.95 0 Td[(9:33)]TJ /F1 11.955 Tf 11.96 0 Td[(0:33f75(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(2:82t3+25:38t2)]TJ /F1 11.955 Tf 11.96 0 Td[(70:61t+279f32(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(3:27t3+31:99t2)]TJ /F1 11.955 Tf 11.95 0 Td[(2:76t)]TJ /F1 11.955 Tf 11.95 0 Td[(81:14f76(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(0:29t2+0:87t+0:05f33(t)=)]TJ /F1 11.955 Tf 9.29 0 Td[(1:06t3+4:47t2)]TJ /F1 11.955 Tf 11.95 0 Td[(3:19t)]TJ /F1 11.955 Tf 11.95 0 Td[(30:32f77(t)=)]TJ /F1 11.955 Tf 9.3 0 Td[(0:22t2+1:03t)]TJ /F1 11.955 Tf 11.95 0 Td[(6:57f34(t)=0f78(t)=0f41(t)=0f85(t)=0f42(t)=0f86(t)=1f43(t)=1f87(t)=0f44(t)=0f88(t)=0(6{4) Altogether,thetwelvestatevectorincludesthethreeorientationangles(,, ),thethreeangularrates(p,q,r),thepositionvector([xyz]),andthevelocityvector([uvw]).Forthepurposesofconvenience,thestatevectorwillbereducedtoeightstatesandaugmentedtoinclude.Thereduced,augmentedstatevectorcanbeshowninequation 6{5 171
PAGE 172
x(t)=2666666666666666666664u(t)w(t)q(t)(t)(t)p(t)r(t)(t)3777777777777777777775(6{5) 6.5.2.2Time-varyingpolesandstabilitymodes Applyingthetime-varyingalgorithmdenedinSection 3.5 tothetime-varyingmodelinequation 6{3 ,ateacht2R+,producesanLTVpoleset,stabilitymodes,andcorrespondingeigenvectors.Duetotheextensiveamountofdataproducedforeachmorphingconguration,aswellasthepotentialenvelopeforwhichtheaircraftcanmorph,onlycertainmorphingtrajectoriesandcongurationsarechosentostudythetime-varyingdynamics.Inadditiontolimitingtheamountofcongurations,onlythelongitudinalsystemdynamicsareconsidered.Thecongurationsarechosensuchthattheyincludedsymmetricsweepfromstraight(0degrees)tosweptpositionsof10,20,and30degrees.Eachofthesemorphingtrajectoriesareperformedatratesof2,5,10,and20seconds.Increasingtherateofmorphingalsoincreasestheoveralleectthetime-varyinginertiacontributesintothesystem'sdynamics. Thesetime-varyingeectsareevident,asshowninFigures 6-56 6-59 .Clearly,thestabilitymodesinFigures 6-56 6-57 showtwomodeswhichconvergeuponzeroquickly(modesAandB),whiletwomodesconvergeuponzeroaftersomelengthyperiodoftime(modesCandD).ObservationsshowthatmodesAandBshowlittlechangefromvaryingthemorphingrateorangle.ThisresultissimplyduetothefactmodesAandBhavealreadyconvergeduponzerobeforeanyinertialeectshavebeenintroduced. 172
PAGE 173
A B C D Figure6-56. Stabilitymodechangevstimechange:10degrees(|),20degrees()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[(),30degrees(---)A)2secondsB)5secondsC)10secondsD)20seconds Conversely,modesCandDdoexperiencetheeectsofinertiaandthereforeshowlargedierencesastimeprogresses.ObservationsofFigures 6-56 6-57 suggestthatastheallowablemorphingtimeisreduced,thedeviationincoincidentmodesshapesincrease.Thisobservationleadstothefollowingintuitiveresult;asthemorphingtimeisreduced,theoveralleectsofinertiaincrease,andthereforeproducelargersystemerrorswhicharepreviouslyunaccountedfor. ThepolesetscorrespondingtotheaforementionedstabilitymodesareshowninFigures 6-58 6-59 .Likethetrendshownwiththestabilitymodes,thepolesetsalsodeviateintimebaseduponthemoprhingrateorangle.Thisobservationisintuitivedue 173
PAGE 174
A B C Figure6-57. Stabilitymodechangevsdegreechange:20seconds(|),10seconds()-203()-204()]TJ /F1 11.955 Tf 32.77 0 Td[(),5seconds(---),2seconds()A)10degreesB)20degreesC)30degrees tothefactthatthepolesetsaredirectlycomputedfromthestabilitymodes.ItshouldbenotedthatinFigures 6-59 A6-59 B,thatsomepolesetsseemtoconvergeuponaconstantvalue.Thisconvergingactionisduetothefactthattherespectivepolesethasreachedtheendofitsmorphingmaneuverandhasindeedbecometime-invariant.Thesystem'stime-invariancedemonstratesthatthestateequationisT-periodicandthereforeavalidrepresentationofthetime-varyingpolesetcanbegivenbythesystem'sfrozentimeeigenvalue[ 86 ].AnotherimportantobservationderivedfromFigures 6-58 6-59 isthattwopolesetscreateahighfrequency,oppositephase,oscillatingpair,whiletheothertwopolesetscreatealowfrequency,oppositephase,oscillatingpair.Recallfromthe 174
PAGE 175
previouslyshownmechanicalsystemexamplethatanoppositephase,oscillatingpolepairisrepresentativeofdiscretecomplex-conjugatepolepair.Usingthisfact,itcanstatedthatthepolesetsgeneratedforthelongitudinalaircraftexamplearedirectlycomparabletotheclassicallydenedtimeinvariantphugoidandshortperioddynamicmodes. A B C D Figure6-58. Polechangevstimechange:10degrees(|),20degrees()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[(),30degrees(---)A)2secondsB)5secondsC)10secondsD)20seconds Togainabetterunderstandingforthedampingandfrequencyelementsoftheselongitudinalaircraftmodes,letusconsiderasinglecasewheretheaircraftmorphs30degreesin10seconds.Thecorrespondingpolesets,stabilitymodes,andorthonormalvectormatrixareshowninFigures 6-60 6-62 175
PAGE 176
A B C Figure6-59. Polechangevsdegreechange:20seconds(|),10seconds()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[(),5seconds(---),2seconds()A)10degreesB)20degreesC)30degrees ThepolesetsfoundinFigure 6-60 Aoscillatedirectlyaboutlineswhicharecontinuousrepresentationsofthefrozentimeeigenvalues.Aconclusion,similartothatfoundwiththemechanicalsystemexample,canbedrawnsuchthattheaverageoftheoppositephase,oscillatingpolesetpairscanbeusedtondthedampingofthedynamicmodalresponses.Figure 6-60 Aalsoillustratesaninitialtransientregionwheretwopolesetsswap.Thisregion,alongwiththeill-conditionedregionfortheshort-period-likepolesetsaredisregarded.ItcanbeseenfromFigure 6-60 A6-60 Bthattheill-conditionedregionfortheshort-period-likepolesetsdirectlycorrelatestothetime 176
PAGE 177
A B Figure6-60. Longitudinalaircraft30degree,10secondmorphA)polesets:polesets1and2()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[(),polesets3and4(---),time-invariantfrozentimeeigenvalues(|)B)stabilitymodes:mode1(|)mode2()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()mode3(---)mode4() thatthecorrespondingstabilitymodesconvergeuponzero.Thisisthesameintuitiveresultthatwasshownearlierinthemechanicalsystemexample. Animportantobservationtakenfromthepreviousmechanicalsystemexample,inSection 4.4 ,wasthatthedynamicmodalfrequenciescouldbestudiedbylookingattheperiodicityofthecolumnvectorsinQ.Thissameobservationcanbeappliedtothelongitudinalaircraftexample.Fromthisexample,itcanbeseenthatcolumnvectors1and2,asseeninFigure 6-61 ,describeresponseswithagreaterperiodicitythanthatfoundincolumnvectors3and4,asshowninFigure 6-62 Therelativedierencesinperiodicityonceagainsuggeststhatthereexistscomparabletraitsbetweenthecalculatedtime-varyingdynamicmodesandtheclassicallydenedtime-invariantphugoidandshort-perioddynamicmodes.Figures 6-62 A6-62 Drepresentthecolumnvectorsthatdescribethehighlydamped,higherfrequencytime-varyingpolesets,andlikethepolesets,thecolumnvectorsbecomeill-conditionedasthecorrespondingstabilitymodesconvergeuponzero.ThisresultisduetothefactthatthemultiplicationofQandRmustlinearlycombinetocreatethesimulatedstateresponses,asshowninFigure 6-63 177
PAGE 178
A B Figure6-61. LongitudinalaircraftQcolumnvectors1and2for30degree,10secondmorph:A)Q11(|)Q21()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()Q31(---)Q41()B)Q12(|)Q22()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()Q32(---)Q42() Thestatesforthisexampleareu(thebodyrelatedx-axisvelocity(m/s)),w(thebodyrelatedz-axisvelocity(m/s)),q(themeasuredpitchrate(deg/s)),and(themeasuredpitchangle(deg)),wheretheirrelativeresponsesareshowninFigures 6-63 A6-63 D.RecallthatthestateresponsesareunforcedresponseswiththeinitialconditionvectordescribedbyEquation 6{6 266666664u0w0q00377777775=26666666415000377777775(6{6) 6.6FeedforwardControl 6.6.1MorphingBasis Thepolynomialbasisusedtoconstrainthemass-springexample,showninSection 5.4.3 ,willalsobeusedasthebasisconstraintforthisexample.Recall,thatanymorphingtrajectoryisrestrictedtoathird-orderpolynomialoftimeandmustliewithinthesetofallpolynomialswithrealcoecients,asseeninEquation 5{11 178
PAGE 179
A B C D Figure6-62. LongitudinalaircraftQcolumnvectors3and4for30degree,10secondmorph:A,B)Q13(|)Q23()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()Q33(---)Q43()C,D)Q14(|)Q24()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()Q34(---)Q44() Thecoecientsofthemorphingtrajectoryareboundedsuchthattheaircraftneversweepsitswingspast30deg.Thehigher-ordertermsinthepolynomialhavethepotentialtodominateastimeincreases;consequently,thesetermsarerestrictedmorethanthelower-orderterms. Theprocedureofdeterminingtheinitialboundsactuallyndsthecoecientssequentiallybyconsideringonlyonecoecienttobeanon-zerovalueatatime.Indongso,thesolutiondeterminesthemaximumcontributionneededbyeachmorphingcoecienttoremainwithin30deg.ThisprocedurecanbeshowninEquation 6{7 179
PAGE 180
A B C D Figure6-63. LongitudinalaircraftstateresponsesA)state1(u)B)state2(w)C)state3(q)D)state4() iti=30 (max(t))ii=0;:::;3(6{7) Also,aconstraintisintroducedsuchthatj(t)j30forallvaluesoftime.Thisconstraintnotesthatthedesiredsignalisassumedtobeboundedandthusthesystemmustexperiencerealisticgeometrychange.Thecoecientsofeachpolynomialareadditionallylimitedtomaintainthisboundedcondition. TheresultingboundsonthecoecientsaregiveninTable 6-10 180
PAGE 181
A B C D Figure6-64. Longitudinalaircraftcolumneigenvectorsfor30degree,10secondmorph:A)V11(|)V21()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()V31(---)V41()B)V12(|)V22()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()V32(---)V42()C)V13(|)V23()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()V33(---)V43()D)V14(|)V24()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()V34(---)V44() Table6-10. Boundsoncoecientofmorphingbasis CoecientUpperBoundLowerBound 030-3012-220.133-0.13330.009-0.009 181
PAGE 182
6.6.2TrackingASystemResponse 6.6.2.1Polynomialmorphing Asetofdesiredresponsesaregeneratedusingpolynomialmorphing.Thesepolynomialsrangeasarst-order,second-orderandthird-orderfunctions.TheoptimizedvaluesofthemorphingcommandsaregeneratedforfeedforwardcontrolandshowninTable 6-11 .TheoptimizedmorphingisessentiallyidenticaltothedesiredmorphingwhichisanticipatedsincethedesiredmorphingcanbeexactlyduplicatedbythemorphingbasisofEquation 5{11 Table6-11. Optimalpolynomialmorphingtotrackdesiredpolynomialmorphing desiredactual 5+1:5t5+1:5t1:8t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:08t21:8t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:08t27)]TJ /F1 11.955 Tf 11.95 0 Td[(0:25t)]TJ /F1 11.955 Tf 11.95 0 Td[(0:1t2+0:008t37)]TJ /F1 11.955 Tf 11.95 0 Td[(0:25t)]TJ /F1 11.955 Tf 11.95 0 Td[(0:1t2+0:008t3 Theoptimizedstateresponses,muchliketheoptimizedmorphingtrajectories,areessentiallyidenticaltothedesiredstateresponses.TheseresponsesareshowninFigure 6-65 Anotherdesiredresponseisattemptedtobetrackedusingthepolynomialmorphing;however,thisdesiredresponseisassociatedwithamorphingcomposedofasinusoidaladdedtoapolynomial.ThisdesiredmorphingcannotbeduplicatedusingtherestrictedbasisofEquation 5{11 sosomeerrormustresultinthetracking.TheoptimizedpolynomialformorphingisgiveninTable 6-12 alongwiththisdesiredmorphing. Table6-12. Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing desiredactual 0:75t+cos(t)0:7874+0:153t)]TJ /F1 11.955 Tf 11.95 0 Td[(0:0819t2+0:0025t3 TheresultingstateresponseisshowninFigure 6-66 asreasonablyclosetothedesireresponse.Someerrorispresentaroundthepeaks;however,thedesiredmorphingliescloseenoughtothetherangeofthemorphingbasissuchthatthefeedforwardcontrolisnearlyabletoprovidethecorrecttracking. 182
PAGE 183
A B C Figure6-65. DesiredResponse(|)andOptimalResponse()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()forA)First-OrderMorphingandB)Second-OrderMorphingandC)Third-OrderMorphing 6.6.2.2Piecewise-polynomialmorphing Apiecewise-polynomialmorphingisagainconsideredtoexpandtherangeofthemorphingbasisseeninEquation 5{11 .Thesameconstraintsplacedonthepiecewise-polynomialtechniqueseeninthemass-springexampleapplyhere.Recall,themorphingbasisisallowedtobearst-orderpolynomialwhosecoecientscanvaryatdiscretepointsintime;whereitisalsorequiredthattheoverallmorphingbecontinuousthroughouttheentireresponse. Applyingthepiecewise-polynomialtechniquetothesinusoidalmorphingresponse,resultsintheoptimizedvaluesgiveninTable 6-13 183
PAGE 184
Figure6-66. DesiredResponse(|)andOptimalResponse()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()forSinusoidalMorphing Table6-13. Optimalpiecewise-polynomialmorphingtotrackdesiredsinusoidalmorphing desiredactual 0:75t+cos(t)1:245)]TJ /F1 11.955 Tf 11.96 0 Td[(0:103t:0t2:973:693+0:852t:2:97
PAGE 185
Figure6-67. DesiredResponse(|)andOptimalResponse()-221()-223()]TJ /F1 11.955 Tf 33.2 0 Td[()forSinusoidalMorphing showninTable 6-14 .Itisagainseenthattheoptimizedmorphingisessentiallyidenticaltothedesiredmorphing. Table6-14. Optimalpolynomialmorphingtotrackdesiredpolynomialmorphing desiredactual 8+0:75t8+0:75t)]TJ /F1 11.955 Tf 9.3 0 Td[(3+1:25t+0:02t2)]TJ /F1 11.955 Tf 9.3 0 Td[(3+1:25t+0:02t2)]TJ /F1 11.955 Tf 9.3 0 Td[(5+2t)]TJ /F1 11.955 Tf 11.95 0 Td[(0:1t2+0:008t3)]TJ /F1 11.955 Tf 9.3 0 Td[(5+2t)]TJ /F1 11.955 Tf 11.96 0 Td[(0:1t2+0:008t3 Theoptimizedpolesignalsareessentiallyidenticaltothedesiredpolesignals.ThesesignalsareshowninFigure 6-68 Anotherdesiredsignalisattemptedtobetrackedusingthepolynomialmorphing.Thisdesiredsignalisagainassociatedwithamorphingcomposedofasinusoidaladdedtoapolynomial.Forsimplicity,itremainsthesamedesiredsinusoidalmorphingasthatusedforthestateresponse.TheoptimizedpolynomialformorphingisgiveninTable 6-15 alongwiththisdesiredmorphing. Table6-15. Optimalpolynomialmorphingtotrackdesiredsinusoidalmorphing desiredactual 0:75t+cos(t)1:007+0:078t+0:0736t2)]TJ /F1 11.955 Tf 11.96 0 Td[(0:0089t3 185
PAGE 186
A B C Figure6-68. A)ResponsetoMorphingTrajectory:Desired(|),Optimized()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()B)ResponsetoMorphingTrajectory:Desired(|),Optimized()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[()C)ResponsetoMorphingTrajectory:Desired(|),Optimized()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() Althoughusingthesamedesiredmorphing,theoptimizedpolynomialinTable 6-15 isdierentfromthatinTable 5-4 .Thisresult,likethatseeninthemass-springexample,issimplyduetotheoptimizationofadierentcostfunction. TheresultingpolesignalisshowninFigure 6-69 .Someerrorispresent;however,thefeedforwardcontrolisagainnearlyabletoprovidethecorrecttracking. 6.6.3.2Piecewise-polynomialmorphing Applyingthepiecewise-polynomialtechniquetothesinusoidalmorphingresponse,resultsintheoptimizedvaluesgiveninTable 6-16 186
PAGE 187
Figure6-69. ResponsetoMorphingTrajectory:Desired(|)andOptimized()-222()-222()]TJ /F1 11.955 Tf 33.21 0 Td[() Table6-16. Optimalpiecewise-polynomialmorphingtotrackdesiredsinusoidalmorphing desiredactual 0:94t+sin(t)1:218)]TJ /F1 11.955 Tf 11.96 0 Td[(0:047t:0t2:979:197)]TJ /F1 11.955 Tf 11.96 0 Td[(0:660t:2:97
PAGE 188
CHAPTER7CONCLUSION Thispaperinvestigatestheeectsofwingsweepontheightcharacteristicsforaminiatureairvehicle.Inparticular,thevariationsadmittedbyamulti-jointmechanismarestudied.Suchamechanismallowsindependentchoiceofinboardandoutboardforeachwing.Thisvehicleisactuallyincorporatingabiologically-inspiredapproachthatnotestherotationsofshouldersandelbowsbygulls.Theassociatedaerodynamicsvaryinbothforceandmomentandthewingsweeprangesacrosssymmetricandasymmetriccongurations.Theresultingightdynamicsdemonstratetheasymmetricsweepisbenecialtomaintainingsensorpointingdespitecrosswinds;consequently,thebiomimeticdesignhasenhancedmissioneectivenessforthisclassofvehicles. Theequationsofmorphingarederivedforamorphingaircraftandshowntoincludetermsassociatedwithasymmetriesandtime-varyinginertias.Acriticaleectoftheseelementstermsisacouplingbetweenthelongitudinalandlateral-directionaldynamics.Essentially,thesetermsindicatethatthetransientbehaviorofmorphingcanbesignicantlydierentthanitssteady-statebehavior.Simulationsoftheclosed-loopbehaviorsforavariablewing-sweepaircraft(bothindivingandturning)indicatetheadverseinuencethatsuchtransientbehaviorcanpotentiallyhaveonmissionperformance.Inbothcases,theresultinginertiasproducedduringthetransientresponseresultedinrespectivelongitudinalandlateraltranslationsthatcausedthevehicletomissitstarget. Thispaperalsopresentsndingsbasedonobservationstakenfromparametercomputations[ 86 ]fortwodierentlineartime-varyingsystems.Comparisonsbetweenthetwotime-varyingexamplesaremade,fromwhichconclusionsaredrawnonhowtosolveforanddescribeasystem'stime-varyingdynamicmodes.Specically,itisshownthatforaparticularmorphingaircraftmaneuver,thelongitudinaltime-varyingparametersresembletheclassicallydenedtime-invariantphugoidandshort-periodcharacteristics. 188
PAGE 189
Thedampingandfrequencyrelatedtothetime-varyingdynamicmodesarefoundtoberelatedtotheaverageoftwooppositephase,oscillatingpolesetsandcolumnvectorperiodicityoftheorthonormalQmatrix,respectively.Itshouldbenotedthatduetothenatureofthedynamicmodesderivedinthispaper,onlyoscillating,complex-conjugatepolesandmodeswerestudied. Lastly,thispaperpresentsndingsbasedonamorphingcontroloftwodierentlineartime-varyingsystems.Comparisonsbetweenthetwotime-varyingexamplesaremade,fromwhichconclusionsaredrawnforthetrackingcapablilityofafeedforwarddesign.Specically,itisshownforparticularaircraftmorphingtrajectories,thatfeedforwardtrackingcanbeachievedondesiredstateresponseandpolevaluesignals.Itisalsoshownthattrackingmaybeperformedusingpiecewise-polynomialtechniques.Currently,feedforwardtrackingispreformedo-lineusingpredenedsignalsasdesiredinput. Futureworkwillconsistofstudyingthelateral-directionalaircraftmodel,fromwhichdenitionsforbothpurelyrealandcomplex-conjugatepolesandmodeswillbediscussed.Also,futureworkwillconsistofimplementingthepreviouslydenedfeedforwardcontrollerinsomeformofmulti-elementfeedbackcontroller.Feedbackwillbeutilizedforpurposesofsignalgeneration,noisecompensationandrobustness. 189
PAGE 190
REFERENCES [1] M.AbdulrahimandR.Lind,"FlightTestingandResponseCharacteristicsofaVariableGull-WingMorphingAircraft,"JournalofAircraft,inreview. [2] M.AbdurahimandR.Lind,"ControlandSimulationofaMulti-RoleMorphingMicroAirVehicle,"JournalofGuidance,ControlandDynamics,inreview. [3] M.Abdulrahim,H.Garcia,andR.Lind,"FlightCharacteristicsofShapingtheMembraneWingofaMicroAirVehicle,"JournalofAircraft,Vol.42,No.1,January-February2005,pp.131-137. [4] A.Y.AganovicandZ.Gajic,"TheSuccessiveApproximationProcedureforFinite-TimeOptimalControlofBilinearSystems,"IEEETransactionsonAu-tomationandControl,Vol.39,No.9,pp.1932-1935,Sept.1994. [5] F.Amato,C.Cosentino,A.S.Fiorillo,andA.Merola,"StabilizationofBilinearSystemsViaLinearState-FeedbackControl,"IEEETransactionsonCircuitsandSystems,Vol.56,No.1,January2009. [6] P.ApkarianandR.J.Adams,"AdvancedGain-SchedulingTechniquesforUncertianSystems,"IEEETransactionsonControlSystemsTechnology,Vol.6,No.1,January1998,pp.21-32. [7] P.ApkarianandP.Gahinet,"AConvexCharacterizationofGain-ScheduledH1Controllers,"IEEETransactionsonAutomaticControl,Vol.40,No.5,May1995,pp.853-864. [8] R.K.ArningandS.Sassen,"FlightControlofMicroAirVehicles,"AIAA-2004-4911. [9] J.Bae,T.M.Seigler,D.J.InmanandI.Lee,"AerodynamicandAeroelasticConsiderationsofaVariable-SpanMorphingWing,"JournalofAircraft,Vol.42,No.2,March-April2005,pp.528-534. [10] B.R.ABarmish,"NecessaryandSucientConditionsforQuadraticStabilizabilityofanUncertainSystem,"JournalofOptimation,Theory,andApplication,Vol.46,1985,pp.399-408. [11] G.BeckerandA.Packard,"RobustPerformanceofLinearParametricallyVaryingSystemsusingParametrically-DependentLinearFeedback,"Systems&ControlLetters,Vol.23,1994,pp.205-215. [12] J.D.BendtsenandK.Trangbaek,"RobustQuasi-LPVControlBasedonNeuralState-SpaceModels,"IEEETransactionsonNeuralNetworks,Vol.13,No.2,March2002,pp.355-368. [13] K.Boothe,K.FitzpatrickandR.Lind,"ControllersforDisturbanceRejectionforaLinearInput-VaryingClassofMorphingAircraft,"AIAAStructures,StructuralDynamicsandMaterialsConference,April2005,AIAA-2005-2374. 190
PAGE 191
[14] J.Bowman,"AordabilityComparisonofCurrentandAdaptiveandMultifunctionalAirVehicleSystems,"AIAA-2003-1713. [15] J.Bowman,B.SandersandT.Weisshar,"EvaluatingtheImpactofMorphingTechnologiesonAircraftPerformance,"AIAA-2002-1631. [16] S.Boyd,L.ElGhaoui,E.FeronandV.Balakrishnan,LinearMatixInequalitiesinSystemandControlTheory,SIAM,Philadelphia,PA,1994. [17] C.Bruni,A.GipillpandG.Koch,"BilinearSystems:AnAppealingClassof'NearlyLinear'SystemsinTheoryandApplication,"IEEETransactionsonAutomaticControl,Vol.19,1974,pp.334-348. [18] C.E.S.Cesnik,H.R.LastandC.A.Martin,"AFrameworkforMorphingCapabilityAssessment,"AIAAStructures,StructuralDynamicsandMaterialsConference,April2004,AIAA-2004-1654. [19] C.E.S.Cesnik,A.ArborandE.L.Brown,"ActiveWarpingControlofaJoined-Wing/TailAirplaneConguration,"AIAAStructures,StructuralDynamicsandMaterialsConference,April2003,AIAA-2003-1715. [20] A.Chakravarthy,D.T.GrantandR.Lind,"Time-VaryingDynamicsofaMicroAirVehiclewithVariable-SweepMorphing,"AIAAGuidance,NavigationandControlConference,Chicago,IL,August2009,AIAA-2009-6304 [21] B.M.Chen,RobustandH1Control,TheSpringerCompany,2000. [22] J.C.CockburnandB.G.Morton,"LinearFractionalRepresentationsofUncertainSystems,"Automatica,Vol.33,No.7,1997,pp.1263-1271. [23] V.ConstanzaandC.ENeuman,"OptimalControlofNon-LinearChmeicalReactorsViaanInitial-ValueHamiltonianProblem,"OptimalControlApplicationsandMethods,Vol.27,pp.41-60,2006. [24] R.F.Curtain,"State-SpaceApproachestoH1ControlForInnite-DinmesionalLinearSystems,"TransactionsoftheInstituteofMeasurementandControl,Vol.13,No.5,1991,pp.253-261 [25] H.D'Angelo,LinearTime-VaryingSystems:AnalysisandSynthesis.Boston,MA:Allyn&Bacon,1970. [26] C.Desoer,andJ.Schulman,"Zerosandpolesofmatrixtransferfunctionsandtheirdynamicalinterpretation,"IEEEtrans.CircuitsSyst.,vol.CAS-21,pp.3-8,Jan.1974. [27] R.C.DorfandR.H.Bishop,ModernControlSystems,10thed.UpperSaddleRiver,NJ:Prentice-Hall,2005 191
PAGE 192
[28] J.C.Doyle,"AnalysisofaFeedbackSystemWithStructuredUncertainties,"IEEEProceedings-D,Vol.129,1982,pp.242-250. [29] J.C.Doyle,K.Glover,P.P.Khargonekar,andB.A.Francis,"StateSpaceSolutionstotheStandardH2andH1ControlProblems,"IEEETransactionsonAutomationandControl,Vol.19,1989,pp.831-847. [30] M.Drela,andH.Youngren,"AVL-AerodynamicAnalysis,TrimCalculation,DynamicStabilityAnalysis,AircraftCongurationDevelopment,"AthenaVortexLattice,v.3.15,http://raphael.mit.edu/avl/. [31] E.L.Duke,R.F.AntoniewiczandK.D.Krambeer,DerivationandDenitionofaLinearAircraftModel,NASA-RP-1207,August1988. [32] M.Ekman,"SuboptimalControlfortheBilinearQuadraticRegulatorProblem:ApplicationtotheActivatedSludgeProcess,"IEEETransactionsonControlSystemsandTechnology,Vol.13,No.1,pp.162-168,January2005. [33] B.EtkinandL.D.Reid,DynamicsofFlightStabilityandControl,3rded.JohnWiley&Sons,Inc.,1996 [34] S.M.Ettinger,M.C.Nechyba,P.G.IfjuandM.Waszak,"Vision-GuidedFlightStabilityandControlforMicroAirVehicles,"ProceedingsoftheIEEEConferenceonIntelligentRobotsandSystems,2002,pp.2134-2140. [35] E.Feron,P.ApkarianandP.Gahinet,"AnalysisandSynthesisofRobustControlSystemsviaParameter-DependentLyapunovFunctions,"IEEETransactionsonAutomaticControl,Vol.41,No.7,July1996,pp.1041-1046. [36] I.Fialho,G.J.Balas,A.K.Packard,J.RenfrowandC.Mullaney,"Gain-ScheduledLateralControloftheF-14AircraftduringPoweredApproachLanding,"JournalofGuidance,ControlandDynamics,Vol.23,No.3,May-June2000,pp.450-458. [37] I.FialhoandG.J.Balas,"RoadAdaptiveActiveSuspensionDesignUsingLinearParameter-VaryingGain-Scheduling,"IEEETransactionsonControlSystemsTechnology,Vol.10,No1,January2002,pp.43-54. [38] A.Filippone,"FlightPerformanceofFixedandRotaryWingAircraft,"AmericanInstituteofAeronauticsandAstronautics,Reston,VA,2006,pp.251-253. [39] G.Floquet,AnnalesScientiquesdeI'EcoleNormaleSuperieure,(2)8,p.49,1879 [40] R.A.FreemanandP.V.Kokotovic,RobustNonlinearControlDesign,Birkhauser,BostonMA,1996. [41] P.Gahinet,P.ApkarianandM.Chilali,"AneParameter-DependentLyapunovFunctionsandRealParametricUncertainty,"IEEETransactionsonAutomaticControl,Vol.41,No.3,March1996,pp.436-442. 192
PAGE 193
[42] P.Gahinet,A.Nemirovski,A.J.LaubandM.Chilali,LMIControlToolbox,TheMathworks,Natick,MA,1995. [43] S.E.GanoandJ.E.Renaud,"OptimizedUnmannedAerialVehiclewithWingMorphingforExtendedRangeandEndurance,"AIAA-2002-5668. [44] F.H.Gern,D.J.Inman,andR.K.Kapania,"StructuralandAeroelasticModelingofGeneralPlanformWingswithMorphingAirfoils,"AIAAJournal,Vol.40,No.4,April2002,pp.628-637. [45] K.GloverandJ.C.Doyle,"State-SpaceFormulaeforAllStabilizingControllersThatSatisfyanH1-NormBoundandRelationstoRiskSensitivity,"SystemandControlLetters,Vol.11,1988,pp.162-172. [46] D.T.Grant,M.AbdulrahimandR.Lind,"FlightDynamicsofaMorphingAircraftutilizingIndependentMultiple-JointWingSweep,"AIAAAtmosphericFlightMechanicsConference,Keystone,CO,August2006,AIAA-2006-6505. [47] Grant,D.T.andLind,R.,"EectsofTime-VaryingInertiasonFlightDynamicsofanAsymmetricMorphingAircraft,"AIAAAtmosphericFlightMechanicsConference,HiltonHead,SC,August2007,AIAA-2007-6487. [48] K.Gu,"H1ControlofSystemsUnderNormBoundedUncertaintiesinAllSystemMatrices,"IEEETransactiononAutomationandControl,Vol.39,1994,pp.1320-1322. [49] P.O.Gutman,"StabilizingControllersforBilinearSystems,"IEEETransactionsonAutomationandControl,Vol.AC26,No.4,pp.917-922,August1981. [50] Y.Heryawan,H.C.Park,N.S.Goo,K.J.Yoon,andY.H.Byun,"DesignandDemonstrationofaSmallExpandableMorphingWing,"ProceedingsofSPIE-Volume5764,May2005,pp.224-231. [51] D.E.Hill,J.R.Baumgarten,andJ.T.Miller,"DynamicSimulationofSpin-StabilizedSpacecraftwithSloshingFluidStores,"JournalofGuidance,ControlandDynamics,Vol.11,No.6,November-December1988,pp.597-599. [52] M.Y.Hsiao,C.H.LiuandS.H.Tsai,"ControllerDesignandStabilizationForAClassofBilinearSystems,"IEEETransactionsonCircuitsandSystems,Vol.56,No.1,January2009. [53] R.A.HornandCRJohnson,MatrixAnalysis.Cambridge,U.K.:CambridgeUniv.Press,1990993,vol.3,pp.519-522. [54] A.Isidori,NonlinearControlSystems,Springer-VerlagInc.,Berlin,1995. [55] A.IsidoriandA.Astol,"DisturbanceAttenuationandH1ControlViaMeasurementFeedbackinNonlinearSystems,"IEEETransactiononAutoma-tionandControl,Vol.37,1992,pp.1283-1293. 193
PAGE 194
[56] H.Jerbi,"GlobalFeedbackStabilizationofaNewClassofBilinearSystems,"Syst.ControlLett.,Vol.42,No.4,pp.313-320,April2001. [57] C.O.Johnston,D.A.Neal,L.D.Wiggins,H.H.Robertshaw,W.H.Mason,andD.J.Inman,"AModeltoComparetheFlightControlEnergyRequirementsofMorphingandConventionallyActuatedWings,"AIAAStructures,StructuralDynamicsandMaterialsConference,April2003,AIAA-2003-1716. [58] S.P.Joshi,Z.Tidwell,W.A.CrossleyandS.Ramakrishnan,"ComparisonofMorphingWingStrategiesBasedUponAircraftPerformanceImpacts,"AIAAStruc-tures,StructuralDynamicsandMaterialsConference,April2004,AIAA-2004-1722. [59] E.W.Kamen,"Thepolesandzeroesofalineartime-varyingsystem,"inLinearAlgebraAppl.,1988,vol.98,pp.263-289. [60] E.W.Kamen,"Ontheinnerandouterpolesandzerosofalinear,time-varyingsystem,"inProc.IEEEConf.DecisionandControl,Austin,TX,Dec.1988,pp.910-914. [61] P.P.Khargonekar,I.R.PetersonandM.A.Rotea,"H1OptimalControlWithStateFeedback,"IEEETransactiononAutomationandControl,Vol.33,1988,pp.786-788. [62] P.P.Khargonekar,I.R.PetersonandK.Zhou,"RobustStabilizationofUncertainLinearSystems:QuadraticStabilizabilityandH1ControlTheory,"IEEETransac-tiononAutomationandControl,Vol.35,1990,pp.356-361. [63] J.J.Kehoe,R.S.Causey,M.AbdulrahimandR.Lind,"WaypointNaviagtionforaMicroAirVehicleusingVision-BasedAttitudeEstimation,"AIAAGuidance,NavigationandControlConference,August2005. [64] J.J.Kehoe,J.Grzywna,R.Causey,R.Lind,M.NechybaandA.Kurdila,"ManeuveringandTrackingforaMicroAirVehicleusingVision-BasedFeedback,"SAEWorldofAviationCongress,November2004. [65] H.K.Khalil,NonlinearSystems,PrenticeHall,NewJersey,1996. [66] N.S.Khot,F.E.Eastep,andR.M.Kolonay,"MethodforEnhancementoftheRollingManeuverofaFlexibleWing,"JournalofAircraft,Vol.34,No.5,September-October1997,pp.673-678. [67] J.KimandN.Koratkar,"EectofUnsteadyBladePitchingMotiononAerodynamicPerformanceofMicrorotorcraft,"JournalofAircraft,Vol.42,No.4,July-August2005,pp.874-881. [68] B.S.Kim,Y.J.Kim,andM.T.Lim,"RobustH1StateFeedbackControlMethodsforBilinearSystems,"IEEEProc.-ControlTheoryAppl.,Vol.152,No.5,Sept.2005. 194
PAGE 195
[69] V.B.KolmanovskiiandN.I.Koroleva,"OptimalControlofSomeBilinearSystemsWithAftereect,"PMMU.S.S.R.,Vol.53,No.2,1989,pp.183-188. [70] S.KwakandR.Yedavalli,"NewModelingandControlDesignTechniquesforSmartDeformableAircraftStructures,"JournalofGuidance,ControlandDynamics,Vol.24,No.4,July-August2001,pp.805-815. [71] S.Lall,K.Glover,,"RobustPerformanceandAdaptationUsingRecedingHorizonH1ControlofTime-VaryingSystems,"ProceedingsoftheAmericanControlConference,1995,pp.2384-2389. [72] B.S.Lazos,"BiologicallyInspiredFixed-WingCongurationStudies,"JournalofAircraft,Vol.42,No.5,September-October2005,pp.1089-1098. [73] L.H.LeeandM.Spillman,"Robust,Reduced-Order,LinearParameter-VaryingFlightControlforanF-16,"AIAAGuidance,NavigationandControlConference,August1997,AIAA-97-3637,pp.1044-1054. [74] T.Liu,K.Kuykendoll,R.RhewandS.Jones,"AvianWings,"AIAA-2004-2186. [75] L.Ljung,SystemIdentication:TheoryfortheUser.EnglewoodClis,NJ:Prentice-Hall,1987 [76] R.Longchamp,"StableFeedbackControlofBilinearSystems,"IEEETransactionsonAutomationandControl,Vol.AC25,No.2,pp.302-306,April1980. [77] M.H.Love,P.S.Zink,R.L.Stroud,D.R.Bye,andC.Chase,"ImpactofActuationConceptsonMorphingAircraftStructures,"AIAA-2004-1724. [78] A.G.J.MacFarlaneandN.Karcanias,"Polesandzerosoflinearmulti-variatesystems:Asurveyofalebraic,geometricandcomplex-variabletheory,"Int.J.Control,vol.24,no.1,pp.33-74,1976. [79] D.McLean,AutomaticFlightControlSystems,ThePrentice-HallCompany,NewYork,NY,1990. [80] T.Melin,AVortexLatticeMATLABImplementationforLinearAerodynamicWingApplications,M.S.Thesis,RoyalInstituteofTechnology,Stockholm,Sweden,2000. [81] R.R.Mohler,NonlinearSystems:ApplicationtoBilinearControl,ThePrentice-HallandEnglewoodClisCompanies,NewJersey,1991. [82] A.NatarajanandH.Schaub,"LinearDynamicsandStabilityAnalysisofaTwo-CraftCoulombTetherFormation,"JournalofGuidance,ControlandDy-namics,Vol.29,No.4,July-August2006,pp.831-838. [83] R.C.Nelson,FlightStabilityandAutomaticControl,TheMcGraw-HillCompanies,Boston,MA,1998. 195
PAGE 196
[84] R.A.Nicols,R.T.ReichertandW.J.Rugh,"GainSchedulingforH-InnityControllers:AFlightControlExample,"IEEETransactionsonControlSystemTechnology,Vol.1,No.2,June1993,pp.69-79. [85] U.M.L.Norberg,"Structure,FormandFunctionofFlightinEngineeringandtheLivingWorld,"JournalofMorphology,No.252,2002,pp.52-81. [86] R.T.O'BrienandP.Iglesias,"Polesandzerosfortime-varyingsystems,"Ph.D.dissertation,Dept.Elect.Comp.eng.,TheJohnsHopkinsUniv.,Baltimore,MD,1997. [87] K.Ogata,ModernControlEngineering,Prentice-Hall,UpperSaddleRiver,NJ,2002,pp.843-909. [88] K.Ogata,SystemDynamics,EnglewoodClis,NJ:Prentice-Hall,1978 [89] R.I.OliverandS.F.Asokanthan,"Control/StructureIntegratedDesignforFlexibleSpacecraftundergoingOn-OrbitManeuvers,"JournalofGuidance,ControlandDynamics,Vol.20,No.2,March-April1997,pp.313-319. [90] A.Packard,"GainSchedulingviaLinearFractionalTransformation,"SystemsandControlLetters,Vol.22,1994,pp.79-92. [91] G.Papageorgiou,K.Glover,G.D'MelloandY.Patel,"DevelopmentofaReliableLPVModelfortheLongitudinalDynamicsofDERAsVAACHarrier,"AIAAGuidance,NavigationandControlConference,August2000,AIAA-2000-4459. [92] I.R.Peterson,"DisturbanceAttentuationandH1Optimization:ADesignMethodBasedonTheAlgebraicRicattiEquation,"IEEETransactiononAutomationandControl,Vol.32,1987,pp.427-429. [93] B.C.Prock,T.A.WeisshaarandW.A.Crossley,"MorphingAirfoilShapeChangeOptimizationwithMinimumActuatorEnergyasanObjective,"AIAA-2002-5401. [94] D.L.RaneyandE.C.Slominski,"MechanizationandControlConceptsforBiologicallyInspiredMicroAirVehicles,"AIAA-2003-5345. [95] A.V.Rao,DynamicsofParticlesandRigidBodies:ASystematicApproach,CambridgeUniv.Press,Cambridge,NY,2006,pp.42-51. [96] J.Roskam,AirplaneFlightDynamicsandAutomaticFlightControls,DARcorporation,Lawrence,KS,2001. [97] W.J.Rugh,LinearSystemTheory,2nded.EnglewoodClis,NJ:Prentice-Hall,1993. [98] W.J.RughandJ.S.Shamma,"ResearchonGainScheduling,"Automatica,Vol.36,2000,pp.1401-1425. 196
PAGE 197
[99] M.T.Rusnell,S.E.Gano,V.M.Perez,J.E.Renaud,andS.M.Batill,"MorphingUAVParetoCurveShiftforEnhancedPerformance,"AIAA-2004-1882. [100] M.Sampei,T.Mita,M.Nakamichi,"AnAlgebraicApproachtoH1OutputFeedbackControlProblems,"SystemsandControlLetters,Vol.14,1990,pp.13-24 [101] B.Sanders,F.E.Eastep,andE.Forster,"AerodynamicandAeroelasticCharacteristicsofWingswithConformalControlSurfacesforMorphingAircraft,"JournalofAircraft,Vol.40,No.1,January-February2003,pp.94-99. [102] B.Sanders,G.Reich,J.JooandF.E.Eastep,"AirVehicleControlUsingMultipleControlSurfaces,"AIAAStructures,StructuralDynamicsandMaterialsConference,April2004,AIAA-2004-1887. [103] M.Secanell,A.SulemanandP.Gamboa,"DesignofaMorphingAirfoilforaLightUnmannedAerialVehicleusingHigh-FidelityAerodynamicShapeOptimization,"AIAA-2005-1891. [104] J.S.ShammaandJ.R.Cloutier,"Gain-ScheduledMissileAutopilotDesignUsingLinearParameterVaryingTransformations,"JournalofGuidance,ControlandDynamics,Vol.16,No.2,March-April1993,pp.256-263. [105] P.Shi,S.P.Shue,Y.ShiandR.K.Agarwal,"ControllerDesignforBilinearSystemswithParametricUncertainties,"MathematicalProblemsinEngineering,Vol.4,1999,pp.505-528. [106] S.P.Shue,"OptimalandH1SuboptimalFeedbackControlofDynamicalNonlinearSystemsforControlofWingRockMotion,"Ph.D.Dissertation,WichitaStateUniversity,Spring1997. [107] W.Shyy,M.BergandD.Ljungqvist,"FlappingandFlexibleWingsforBiologicalandMicroAirVehicles,"ProgressinAerospaceSciences,Vol.35,No.5,1999,pp.455-506. [108] K.C.Slatton,W.E.Carter,R.L.ShresthaandW.Dietrich(2007),AirborneLaserSwathMapping:Achievingtheresolutionandaccuracyrequiredforgeosurcialresearch,Geophys.Res.Lett.,34,L23S10,doi:10.1029/2007GL031939. [109] J.SlotineandW.Li,AppliedNonlinearControl,PrenticeHall,1990. [110] R.SmithandA.Ahmed,"RobustParametricallyVaryingAttitudeControllerDesignsfortheX-33Vehicle,"AIAAGuidance,NavigationandControlConference,August2000,AIAA-2000-4158. [111] E.D.Sontag,"ALyapunov-LikeCharacterizationofAsymptoticControllability,"SIAMJournalonControlandOptimization,Vol.21,1983,pp.462-471. [112] S.S.Sorley,"IdenticationandAnalysisofTime-VaryingModalParamters,"UniversityofFloridaMaster'sThesis,2010 197
PAGE 198
[113] A.G.Sparks,"LinearParameterVaryingControlforaTaillessAircraft,"AIAAGuidance,NavigationandControlConference,August1997,AIAA-97-3636,pp.1035-1043. [114] C.H.SpennyandT.E.Williams,"LibrationalInstabilityofRigidSpaceStationduetoTranslationofInternalMass,"JournalofGuidance,ControlandDynamics,Vol.14,No.1,January-February1991,pp.31-35. [115] B.L.StevensandF.L.Lewis,AircraftControlandSimulation,JohnWileyandSons,Hoboken,NJ,2003,pp.308-338. [116] W.Tan,A.K.PackardandG.J.Balas,"Quasi-LPVModelingandLPVControlofaGenericMissile,"AmericanControlConference,2000,pp.3692-3696. [117] G.Tao,S.Chen,J.FeiandS.M.Joshi,"AnAdaptiveActuatorFailureCompensationSchemeforControllingaMorphingAircraftModel,"IEEECon-ferenceonDecisionandControl,December2003,pp.4926-4931. [118] S.W.ThurmanandH.Flashner,"RobustDigitalAutopilotDesignforSpacecraftEquippedwithPulse-OperatedThrusters,"JournalofGuidance,ControlandDynamics,Vol.19,No.5,September-October1996,pp.1047-1055. [119] H.D.Tuan,E.Ono,P.ApkarianandS.Hosoe,"NonlinearH1ControlforanIntegratedSuspensionSystemviaParametrizedLinearMatrixInequalityCharacterizations,"IEEETransactionsonControlSystemsTechnology,Vol.9,No.1,January2001,pp.175-185. [120] S.VenkataramananandA.Dogan,"DynamicEectsofTrailingVortexwithTurbulenceandTime-VaryingInertiainAerialRefueling,"AIAA-2004-4945. [121] M.S.VestandJ.Katz,"AerodynamicStudyofaFlapping-WingMicroUAV,"AIAA-99-0994. [122] R.Wall,"TakingShape,"AviationWeekandSpaceTechnology,January5,2004,pp.54-56. [123] R.Wall,"DarpaEyesMaterialsForMorphingAircraft,"AviationWeekandSpaceTechnology,January5,2004,pp.54-56. [124] B.Wie,"SolarSailAttitudeControlandDynamics,Part2,"JournalofGuidance,ControlandDynamics,Vol.27,No.4,July-August2004,pp.536-544. [125] J.R.Wilson,"MorphingUAVsChangeShapeofWarfare,"AerospaceAmerica,February2004,pp.28-29 [126] M.Y.Wu,"Onstabilityoflineartime-varyingsystems,"Int.J.Systs.Sci.,vol.15,no.2,pp.137-150,1984. 198
PAGE 199
[127] M.Y.Wu,"Anewconceptofeigenvaluesandeigenvectorsanditsapplications,"IEEETrans.Automat.Contr.,vol.AC-25,pp.824-826,May1980. [128] J.YuandA.Sideris,"H1ControlWithParametricLyapunovFunctions,"SystemsandControlLetters,Vol.30,pp.57-69,1997. [129] K.Zhou,andP.Khargonekar,"AnAlgebriacRicattiEquationApproachtoH1Optimization,"SystemsandControlLetters,Vol.11,pp.85-91,1988. [130] J.J.Zhu,"PD-spectraltheoryformultivariablelineartime-varyingsystems,"inProc.IEEEConf.DecisionandControl,SanDiego,CA,Dec.1197,pp.3908-3913. [131] J.J.ZhuandC.D.Johnson,"Uniedcanonicalformsformatricesoveradierentialring,"LinearAlgebraAppl.,vol.147,pp.201-248,1991. 199
PAGE 200
BIOGRAPHICALSKETCH DanielThurmondGrantwasborninGainesville,Floridain1983.Asachild,heoftenfoundhimselfintriguedbythemechanicaloperationofremote-controlledtoys.Whennotlostinapileofnutsandbolts,hefoundenjoymentinbuildingmodelairplaneswithhisfather.Thecombinationofthesetwohobbieseventuallyledhimtodevelopastrongpassionforaeronauticalengineering.DanielgraduatedfromSantaFeHighSchoolin2002,afterwhichheenrolledattheUniversityofFlorida.WhileanundergraduateattheUniversityofFlorida,DanieljoinedtheMicro-AerialVehicle(MAV)Laboratory,wherehelearnedhowtobuildandysmallUnmannedAerialVehicles(UAVs).Thishands-onexperienceallowedDanieltoco-leadaDesign,Build,Fly(DBF)teamthatplacedninthataninternationalAIAAstudentcompetitionheldinWichita,Kansas.Asasenior,DanieljoinedtheFlightControlLabattheUniversityofFlorida,wherehedesigned,builtandewamulti-jointed,wing-sweepingMAV.Upongraduatinginthefallof2006,DanielenrolledintograduateschoolattheUniversityofFlorida.Danielreceivedhismaster'sdegreeinthespringof2009andthenhisdoctorateinthespringof2011.BothdegreesDanielearnedwereinaerospaceengineeringwithaprimaryfocusonadvancedightcontrolsanddynamics.Throughouthiscollegiatecareer,Danielexperiencedamultitudeofhighlightingevents.Theseeventsincludedbutnotlimitedto: WinningthetheAtmosphericFlightMechanicsstudentbestpaperawardatthe2006Guidance,Navigation,andControlConferenceinKeystone,Colorado TravelingtoWashingtonD.C.topresenthisworkforCongressduringtheeveningofaPresidentialStateoftheUnionAddress ObtainingaU.S.patentforhisworkonmorphingandurbancamouage ReceivingtheopportunitytoincludehisworkinandauthorabookchapterpublishedfortheAIAAeducationalseries FeaturinghisworkinnumerousdocumenturiesincludingNationalGeogrpahic,theScienceChannel,andSirDavidAttenborough'sim,"FlyingMonsters3D" 200
|
|