位置伺服系统PID控制策略研究与应用

杭州电子科技大学硕士学位论文位置伺服系统PID控制策略研究与应用姓名：禹牛云申请学位级别：硕士专业：检测技术与自动化装置指导教师：薛凌云20100101ILEDLEDLEDLEDLEDTakahashiPIDPIDJacobianPIDPIDPIDPIDPIDPIDPIDPIDPIDPIDLEDPIDPIDPIDIIPIDPIDPIDIIIABSTRACTLEDdietestingandsortingsystemincludesanumberofpositionservosystems,high-performancepositionservosystemisoneofthekeytechnologiesinthesystem.DuetothesmallsizeandpitchintervalofLEDdies,highdynamicandstaticperformanceofpositionservosystemsarerequired,thatisthepositionservosystemshavethecharacteristicsofhighpositioningaccuracy,fastresponse,largeadjustmentrangeandhighanti-jammingability.Positionservosystemisanonlineartime-varyingsystemwithstrongcoupling,avarietyofuncertainties.It’sdifficulttoacquireanaccuratemathematicalmodel.Tothebasicrequirementsandmaincontrolproblemsofpositionservosysteminpracticalapplication,thetheoreticalanalysisaboutthesystemstructureiscarriedout,andthesystematiccontrolschemeisestablished.Researchresultswillbeappliedtopracticalengineeringapplicationsinordertoobtaingoodcontroleffect.Thispapermainlydescribesthemechanismmodelanalysisofpositionservosystem,mathematicalmodelingandtheanalysisanddesignofrelevantcontrolstrategy.SomeofthesecontrolstrategyisappliedtocontrolthepositionservosystemofLEDdietestingandsortingsystem,andsomeapplicationresultsareobtained.Theoperationcharacteristicandprincipleofpermanentmagnetsynchronousmotorisanalyzed,anditsmathematicalmodelisdiscussed.Becausepermanentmagnetsynchronousmotorisnon-linear,strongcoupling,motormodelisdecoupledwiththeprincipleofvectorcontrol.Then,thethree-loopstructureandmechanismisanalyzed,whichisconsistedbycurrentloop,velocityloop,positionloop.Takingasecond-ordersystemasthestudyobject,whichiscommoninpositionservosystem.Themodelparametersofsecond-ordersystemareidentifiedbyrecursiveleastsquarealgorithm,andthentheTakahashimethodisusedtoobtaintheinitialPIDcontrolparameters.BecausetheconventionalPIDcontrolhasthedefectsoflargeovershoot,poorstability,JacobianisusedtooptimizethePIDcontrolparameters,whichisprovidedbyneuralnetworkidentifier.FuzzyPIDcontrolisanimportantresearchdirectionofintelligentcontrol.Analyzingthephaseplaneofsystemstepresponse,anddevelopingfuzzyPIDcontrolrulesfromexpertiseistoreduceexpertisedependence.Thesimulationresultindicatesthat:ComparedwithconventionalPIDcontrol,fuzzyPIDcontrolcanreduceovershootandimprovesystemsteadyperformance.Whenthecontrolsystemrunsinlargedynamicrange,conventionalfuzzyPIDcontrolisshortofself-adjustingcapabilityoftheuniverseandfuzzyrules,anditwillaffectthesystemcontrolperformance.Therefore,introducingthevariableuniverse,andauniverseself-adjustingIVfuzzyPIDcontrolisdesignedonfuzzyrules.Thesimulationresultindicatesthat:VariableuniversefuzzyPIDhastheadvantageofsmallovershoot,strongdynamicadjustingability.Neuralnetworkhasself-learningandself-adjustingability,whichcanusedtooptimizeandadjusttheparametersofuniverseself-adjustingmechanism.Anditwillsimplifythedesignofuniverseself-adjustingmechanism.Thesimulationresultindicatesthat:Whenthefuzzypartitionandrulesaren’treasonable,theparametersofvariableuniversefuzzyPIDcontrolcanbeoptimizedandadjustedaftersometimeneuralnetworklearningandtraining,whichenhancesthecontrolperformanceinsomedegree.UsingattenuationratiomethodistogettheinitialPIDcontrolparameters,whichisappliedinpositionservosystemofLEDdietestingandsortingsystem.Analyzingitsapplicationresult,ithasthedefectionofovershoot,longssttlingtime.Tosolvetheseproblems,neuralnetworkisusedtoadjustandoptimizethePIDparametes.Theexperimentalresultshowsthat:afterPIDparametersoptimizedbyneuralnetwork,theovershootofsystemandsettlingtimeisdecreased.Thesystemperformanceisimproved.Keywords:vectorcontrol,systemidentification,PID,variableuniversefuzzyPID,fuzzyneuralnetwork11.11.1.1[1]8080PMSM[2][3]2DSP1.1.2[4]1.1dθeθuθ+−1.1PCdθθ-edθθθ=ueθ31.2[5][6]PIDPID[7]PIDPID[8][6][9-11][12][13]4[14][15]PID1.3PIDPIDPIDPIDPIDPIDPIDPIDPID[16]PIDPIDPID[17]PIDPIDPIDPID[18]PIDPIDPID[19][20]PIDPIDPIDPID[21-23]PIDPIDPIDPIDPIDPIDPIDPID5PIDPIDPID[24-26]PIDPIDPIDPIDPIDPIDPIDPIDPID[27-28]1.4LEDLEDLEDLEDPID[29][30]PIDPIDPIDPID[31][32]PID61.2PIDinroutyceeceαeαpK∆iK∆dK∆dedtecepidβ1.2PIDPMSMPMSMPIDPIDPIDPIDPIDPIDPIDPIDLEDPIDPID1.57PIDPIDPIDPIDece8LED0di=2.1Blaschke1971[1][4]2.1.1abcdq2.12.2OabcNSfϕθOaNSθdqSiα2.1abc2.2dq2.1OaObOcabcabcaθ2.2d90oqadθdq9abcdq0000coscos(120)cos(120)sinsin(120)sin(120)adbqciiiiiθθθθθθ−+=⋅−−−−+(2.1)0000sinsin(120)sin(120)23coscos(120)cos(120)adbqcuuuuuθθθθθθ−+=⋅−+(2.2)(2.1)aibicidiqidq(2.2)aubucuduqudq2.1.2[1][4]120odqPWSMqdddnrqdddqqfnrdqnrdqqqqLdiuRipidtLLLdiupLRipidtLLLLωϕωω=−+=−−−(2.3)(2.3)dLqLdqfϕRrωnpqqqdddffmdfLiLiLiϕϕϕϕ==+=(2.4)(2.4)dϕqϕdqfi33()[()]22endqqdnfqqddqTpiipiLLiiϕϕϕ=−=−−(2.5)10rerLdJTBTdtωω=−−(2.6)(2.6)LTeTB2.1.3dqqsmqdsmdLLLLLLσσ=+=+(2.7)(2.7)sLσdqmdLmqLdq2.3()()()ddddmdfqqqdqqddmdfduRiLiLiLidtduRiLiLiLidtωω=++−=+−+(2.8)(2.8)nrpωω=REsLσmdLfiqqLiωduREsLσmdL()ddmdfLiLiω+qu(a)d(b)q2.3dq2.1.40di=cos1ϕ=/0di=d0qqqdfLiϕϕϕ==(2.9)32enfqTpiϕ=(2.10)(2.3)(2.6)///3//2nfqqqrLnfrRLpLiuLiTJ