徐州工程學(xué)院畢業(yè)設(shè)計(jì)外文翻譯學(xué)生姓名學(xué)院名稱機(jī)電工程學(xué)院專業(yè)名稱機(jī)械設(shè)計(jì)制造及其自動(dòng)化指導(dǎo)教師2011年5月27日英文原文AdaptiverobustposturecontrolofaparallelmanipulatordrivenbypneumaticmusclesKeywords:PneumaticmuscleParallelmanipulatorAdaptivecontrolNonlinearrobustcontrolAbstractRathersevereparametricuncertaintiesanduncertainnonlinearitiesexistinthedynamicmodelingofaparallelmanipulatordrivenbypneumaticmuscles.Thoseuncertaintiesnotonlycomefromthetime-varyingfrictionforcesandthestaticforcemodelingerrorsofpneumaticmusclesbutalsofromtheinherentcomplexnonlinearitiesandunknowndisturbancesoftheparallelmanipulator.Inthispaper,adiscontinuousprojection-basedadaptiverobustcontrolstrategyisadoptedtocompensateforboththeparametricuncertaintiesanduncertainnonlinearitiesofathree-pneumatic-muscles-drivenparallelmanipulatortoachievepreciseposturetrajectorytrackingcontrol.Theresultingcontrollereffectivelyhandlestheeffectsofvariousparametervariationsandthehard-to-modelnonlinearitiessuchasthefrictionforcesofthepneumaticmuscles.Simulationandexperimentalresultsareobtainedtoillustratetheeffectivenessoftheproposedadaptiverobustcontroller.2008ElsevierLtd.Allrightsreserved.1.IntroductionPneumaticmuscleisanewtypeofflexibleactuatorsimilartohumanmuscle.Itisusuallymadeupofarubbertubeandcrossweavesheathmaterial.Pneumaticmuscleshavetheadvantagesofcleanness,cheapness,light-weight,easymaintenance,andthehigherpower/weightandpower/volumeratioswhencomparedwithpneumaticcylinders(Ahn,Thanh,&Yang,2004).Thebasicworkingprincipleofapneumaticmuscleisasfollows:whentherubbertubeisinflated,thecross-weavesheathexperienceslateralexpansion,resultinginaxialcontractiveforceandthemovementoftheendpointpositionofthepneumaticmuscle.Thus,thepositionorforcecontrolofapneumaticmusclealongitsaxialdirectioncanberealizedbyregulatingtheinnerpressureofitsrubbertube.Theparallelmanipulatordrivenbypneumaticmuscles(PMDPM)studiedinthispaperisanewapplicationofpneumaticmuscles.ItconsistsofthreepneumaticmusclesconnectingthemovingarmoftheparallelmanipulatortoitsbaseplatformasshowninFig.1.Bycontrollingthelengthsofthreepneumaticmuscles,threedegrees-of-freedom(DOF)rotationmotionoftheparallelmanipulatorcanberealized.Suchaparallelmanipulatorcombinestheadvantagesofcompactstructureofparallelmechanismswiththeadjustablestiffnessandhighpower/volumeratioofpneumaticmuscles,whichwillhavepromisingwideapplicationsinrobotics,industrialautomation,andbionicdevices.Manyresearchershaveinvestigatedtheprecisepositioncontrolofpneumaticmusclesduringthepastseveralyears.Mostofthemdealtwiththecontrolofsingleorantagonisticpneumaticmuscles.Specifically,Bowler,Caldwell,andMedrano-Cerda(1996),employedanadaptivepole-placementschemetocontrolabi-directionalpneumaticmuscleactuatorsystem,foruseona7-DOFanthropomorphicrobotarm.CaiandYamaura(1996)presentedaslidingmodecontrollerforamanipulatordrivenbyartificialmuscleactuators.Kimura,Hara,Fujita,andKagawa(1997),appliedthefeedbacklinearizationmethodtothepositioncontrolofasingle-inputpneumaticsystemwithathird-orderdynamicsincludingthepressuredynamics.KimotoandIto(2003)addednonlinearrobustcompensationstoalinearcontrollerinordertostabilizethesystemgloballyandachieverobustnesstouncertainnonlinearities.Carbonell,Jiang,andRepperger(2001),Chan,Lilly,Repperger,andBerlin(2003),Repperger,Johnson,andPhillips(1998)andRepperger,Phillips,andKrier(1999),proposedseveralmethodssuchasfuzzybackstepping,gain-scheduling,variablestructureandfuzzyPD+IforaSISOpneumaticmusclesystemwithasecond-orderdynamicstoachieveasymptoticpositiontracking.Lilly(2003),LillyandQuesada(2004)andLillyandYang(2005),appliedtheslidingmodecontroltechniquewithboundarylayertocontrolpneumaticmuscleactuatorsarrangedinbicepandtricepconfigurations.Tian,Ding,Yang,andLin(2004),adoptedtheRPEalgorithmtotrainneuralnetworksformodelingandcontrollinganartificialmusclesystem.Hildebrandt,Sawodny,Neumann,andHartmann(2002)；Sawodny,Neumann,andHartmann(2005),presentedacascadecontrollerforatwoaxisplanararticulatedroboticarmdrivenbyfourpneumaticmuscles.Asreviewedabove,someoftheresearchersdesignedrobustcontrollerswithoutconsideringthepressuredynamics,whiletheeffectofpressuredynamicsisessentialfortheprecisecontrolofpneumaticmuscles(Carbonelletal.,2001；Chanetal.,2003；Lilly,2003；Lilly&Quesada,2004；Lilly&Yang,2005；Reppergeretal.,1998,1999).Someoftheresearchersdevelopedcontrollerswiththeassumptionthatthesystemmodelisaccurate,orthatmodeluncertaintiessatisfymatchingconditiononly,whilethoseassumptionsarehardtobesatisfiedinpractice(Hildebrandtetal.,2002,2005；Kimuraetal.,1997).ForthePMDPMshowninFig.1,itnotonlyhasallthecontroldifficultiesassociatedwiththepneumaticmuscles,butalsotheaddeddifficultiesofthecoupledmulti-input-multi-output(MIMO)complexdynamicsoftheparallelmanipulatorandthelargeextentofunmatchedmodeluncertaintiesofthecombinedoverallsystem.Inotherwords,thereexistrathersevereparametricuncertaintiesanduncertainnonlinearities,whicharecausednotonlybythetime-varyingfrictionforcesandstaticforcemodelingerrorsofpneumaticmusclesbutalsobytheinherentcomplexnonlinearitiesandunknowndisturbancesoftheparallelmanipulator.Therefore,itisverydifficulttocontrolpreciselythepostureofthePMDPM.Therecentlyproposedadaptiverobustcontrol(ARC)hasbeenshowntobeaveryeffectivecontrolstrategyforsystemswithbothparametricuncertaintiesanduncertainnonlinearities(Xu&Yao,2001；Yao,2003；Yao,Bu,Reedy,&Chiu,2000；Yao&Tomizuka,2001).Thisapproacheffectivelyintegratesadaptivecontrolwithrobustcontrolthroughutilizingon-lineparameteradaptationtoreducethee