会员注册 | 登录 | 微信快捷登录 QQ登录 微博登录 | 帮助中心 人人文库renrendoc.com美如初恋!
站内搜索 百度文库

热门搜索: 直缝焊接机 矿井提升机 循环球式转向器图纸 机器人手爪发展史 管道机器人dwg 动平衡试验台设计

   首页 人人文库网 > 资源分类 > PDF文档下载

外文翻译--混合位置力控制SCORBOT-ER4支机械手与神经网络的非线性补偿 英文版.pdf

  • 资源星级:
  • 资源大小:319.06KB   全文页数:9页
  • 资源格式: PDF        下载权限:注册会员
您还没有登陆,请先登录。登陆后即可下载此文档。
  合作网站登录: 微信快捷登录 支付宝快捷登录   QQ登录   微博登录
友情提示
2:本站资源不支持迅雷下载,请使用浏览器直接下载(不支持QQ浏览器)
3:本站资源下载后的文档和图纸-无水印,预览文档经过压缩,下载后原文更清晰   

外文翻译--混合位置力控制SCORBOT-ER4支机械手与神经网络的非线性补偿 英文版.pdf

HybridPosition/ForceControloftheSCORBOTER4pcManipulatorwithNeuralCompensationofNonlinearitiesPiotrGierlakRzeszowUniversityofTechnology,DepartmentofAppliedMechanicsandRobotics8Powsta´nc´owWarszawySt.,35959Rzesz´ow,Polandpgierlakprz.edu.plAbstract.Theproblemofthemanipulatorhybridposition/forcecontrolisnottrivialbecausethemanipulatorisanonlinearobject,whoseparametersmaybeunknown,variableandtheworkingconditionsarechangeable.Theneuralcontrolsystemenablesthemanipulatortobehavecorrectly,evenifthemathematicalmodelofthecontrolobjectisunknown.Inthispaper,thehybridposition/forcecontrollerwithaneuralcompensationofnonlinearitiesfortheSCORBOTER4pcroboticmanipulatorispresented.ThepresentedcontrollawandadaptivelawguaranteepracticalstabilityoftheclosedloopsysteminthesenseofLyapunov.Theresultsofanumericalsimulationarepresented.KeywordsNeuralNetworks,RoboticManipulator,TrackingControl,ForceControl.1IntroductionRoboticmanipulatorsaredeviceswhichfinddifferentapplicationsinmanydomainsoftheeconomy.Therequirementsinrelationtoprecisionofmotionandautonomyofmanipulatorsareincreasingaswellasthetasksperformedbythemaremoreandmorecomplex.Incontemporaryindustrialapplicationsitisdesiredforthemanipulatortoexertspecifiedforcesandmovealongaprescribedpath.Manipulatorsareobjectswithnonlinearanduncertaindynamics,withunknownandvariableparametersmasses,massmomentsofinertia,frictioncoefficients,whichoperateinchangeableconditions.Controlofsuchcomplexsystemsisveryproblematic.Thecontrolsystemhastogeneratesuchcontrolsignalsthatwillguaranteetheexecutionofmovementalongapathwithasuitableforceandwithdesiredprecisioninspiteofthechangeableoperatingconditions.Inthecontrolsystemsofindustrialmanipulators,thecomputedtorquemethod1,2fornonlinearitycompensationisused.However,theseapproachesrequirepreciseknowledgeaboutthemathematicalmodelthestructureofmotionequationswithcoefficientsofthecontrolobject.Moreover,insuchanapproach,parametersinthecompensatorhavenominalvaluessothecontrolL.Rutkowskietal.Eds.ICAISC2012,PartII,LNCS7268,pp.433–441,2012.c©SpringerVerlagBerlinHeidelberg2012434P.Gierlaksystemactswithouttakingintoaccountthechangeableoperatingconditions.Intheliteratureexistsmanyvariationofalgorithms,inwhichparametersofthemathematicalmodelofmanipulatorareadapted1,2.Howevertheseapproachesdonoteliminatetheproblemwithstructuraluncertaintyofthemodel.Inconnectionwiththepresentdifficulties,neuralcontroltechniquesweredeveloped3,4,5,6.Inthesemethodsthemathematicalmodelisunnecessary.Thesetechniquesareusedinhybridposition/forcecontroller.Inworks7,8suchcontrollershavebeenpresented.Butinthefirstoftheworksonlyforcenormaltothecontactsurfaceistakingintoaccount,andinthesecondworksomeassumptionishardtosatisfyinpracticalapplications,namelysomestiffnessmatrixwhichcharacterizesfeaturesofenvironmentandallowstocalculatecontactforces,mustbeknown.Inpreviousauthorspaperonlypositioncontrollershavebeenconsidered.Inpresentpaperhybridposition/forceneuralcontrollerisshown.Thisapproachtakesintoaccountallforces/momentswhichactsontheendeffector.Theseforces/momentsaremeasuredbysensorlocatedintheendeffector.2DescriptionoftheSCORBOTER4pcRoboticManipulatorTheSCORBOTER4pcroboticmanipulatorispresentedinFig.1.Itisdrivenbydirectcurrentmotorswithgearsandopticalencoders.Themanipulatorhas5rotationalkinematicpairsthearmofthemanipulatorhas3degreesoffreedomwhereasthegripperhas2degrees.aA1q3yzxOBCOOOd1OAlABlBClCDd1235q1q223u2u1u3q4D4u4q5u5bcontactsurfacec108FEFig.1.aSCORBOTER4pcroboticmanipulator,bschemeThetransformationfromjointspacetoCartesianspaceisgivenbythefollowingequationykq,1HybridPosition/ForceControloftheSCORBOTER4pcManipulator435whereq∈Rnisavectorofgeneralizedcoordinatesanglesofrotationoflinks,kqisakinematicsfunction,y∈Rmisavectorofaposition/orientationoftheendeffectorpointD.Dynamicalequationsofmotionoftheanalysedmodelareinthefollowingform7,9Mq¨qCq,˙q˙qF˙qGqτdtuJThqλτF,2whereMq∈Rnxnisaninertiamatrix,Cq,˙q∈RnisavectorofcentrifugalandCoriolisforces/moments,F˙q∈Rnisafrictionvector,Gq∈Rnisagravityvector,τdt∈Rnisavectorofdisturbancesboundedby||τd||0,u∈Rnisacontrolinputvector,Jhq∈Rm1xnisaJacobianmatrixassociatedwiththecontactsurfacegeometry,λ∈Rm1isavectorofconstrainingforcesexertednormallyonthecontactsurfaceLagrangemultiplier,τF∈Rnisavectorofforces/momentsinjoints,whichcomefromforces/momentsFE∈Rmappliedtotheendeffectorexcepttheconstrainingforces.ThevectorτFisgivenbyτFJbTqFE,3whereJbq∈RmxnisageometricJacobianinbody2.TheJacobianmatrixJhqcanbecalculatedinthefollowingwayJhq∂hq∂q,4wherehq0isanequationoftheholonomicconstraint,whichdescribesthecontactsurface.Thisequationreducesthenumberofdegreesoffreedomton1n−m1,sotheanalysedsystemcanbedescribedbythereducedpositionvariableθ1∈Rn17.Theremainderofvariablesdependonθ1inthefollowingwayθ2γθ1,5whereθ2∈Rm1,andγarisefromtheholonomicconstraint.ThevectorofgeneralizedcoordinatesmaybewrittenasqθT1θT2T.LetdefinetheextendedJacobian7Lθ1bracketleftbiggIn1∂γ∂θ1bracketrightbigg,6whereIn1∈Rn1xn1isanidentitymatrix.Thisallowstowritetherelations˙qLθ1˙θ1,7¨qLθ1¨θ1˙Lθ1˙θ1,8andwriteareducedorderdynamicsintermsofθ1,asMθ1Lθ1¨θ1V1θ1,˙θ1˙θ1F˙θ1Gθ1τdtuJThθ1λJbTθ1FE,9whereV1θ1,˙θ1Mθ1˙Lθ1Cθ1,˙θ1Lθ1.Premultiplyingeq.9byLTθ1andtakingintoaccountthatJhθ1Lθ10,thereducedorderdynamicsisgivenbyM¨θ1V1˙θ1FGτdLTu,10whereMLTML,V1LTV1,FLTF,GLTG,τdLTbracketleftbigτd−JbTFEbracketrightbig.436P.Gierlak3NeuralNetworkHybridControlTheaimofahybridposition/forcecontrolistofollowadesiredtrajectoryofmotionθ1d∈Rn1,andexertdesiredcontactforceλd∈Rm1normallytothesurface.Bydefiningamotionerroreθ,afilteredmotionerrors,aforceerror˜λandanauxiliarysignalυ1aseθθ1d−θ1,11s˙eθΛeθ,12˜λλd−λ,13υ1˙θ1dΛeθ,14whereΛisapositivediagonaldesignmatrix,thedynamicequation10maybewrittenintermsofthefilteredmotionerrorasM˙s−V1sLTfxLTbracketleftbigτd−JbTFEbracketrightbig−LTu,15withanonlinearfunctionfxML˙υ1V1υ1FG,16wherexbracketleftBigeTθ˙eTθθT1dv˙θT1d¨θT1dbracketrightBigT.Themathematicalstructureofhybridposition/forcecontrollerhasaformof7uˆfxKDLs−JThbracketleftBigλdKF˜λbracketrightBig−ν,17whereKDandKFarepositivedefinitematrixesofpositionandforcegain,νisarobustifyingterm,ˆfxapproximatesthefunction16.Thisfunctionmaybeapproximatedbytheneuralnetwork.InthisworkatypicalfeedforwardneuralnetworkFig.2bwithonehiddenlayerisassumed.Thehiddenlayerwithsigmoidalneurons,isconnectedwithaninputlayerbyweightscollectedinamatrixD,andwithanoutputlayerbyweightscollectedinamatrixW.Theinputweightsarerandomlychosenandconstant,buttheoutputweightsinitiallyareequalzero,andwillbetunedduringadaptationprocess.Suchneuralnetworkislinearintheweights,andhasthefollowingdescription3,4fxWTϕxε,18withoutputfromhiddenlayerϕxSDTx,wherexisaninputvector,S.isavectorofneuronactivationfunctions,εisanestimationerrorboundedby||ε||0.ThematrixWisunknown,soanestimationˆWisused,andamathematicaldescriptionofarealneuralnetwork,whichapproximatesfunctionfxisgivenbyˆfxˆWTϕx.19

注意事项

本文(外文翻译--混合位置力控制SCORBOT-ER4支机械手与神经网络的非线性补偿 英文版.pdf)为本站会员(英文资料库)主动上传,人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对上载内容本身不做任何修改或编辑。 若此文所含内容侵犯了您的版权或隐私,请立即通知人人文库网([email protected]),我们立即给予删除!

温馨提示:如果因为网速或其他原因下载失败请重新下载,重复下载不扣分。

copyright@ 2015-2017 人人文库网网站版权所有
苏ICP备12009002号-5