1、1 中文 3150 字 An Optimal Fuzzy-PIControllerforthe High-PerformanceSpeed Control of a PMSM Abstract Thepurposeofthispaperistopresentanadaptivemethodfor improvingthecontrolperformanceofpermanentmagneticsynchronousmotor(PMSM)inoperatingcondition.Theapproachallowstoreducespeedtrackingerrorandtocopewithext
2、ernaldisturbance.Themethodologyofspeedcontrolispresentedindetailandtwocontrollersaretested,traditionalproportionalintegrative(PI)controllerandfuzzyproportionalintegrative(fuzzy-PI)controller.Bothcontrollersshowedgoodresultsfromexperimentspresentingsimilarbehaviors.However,thefuzzy-PIstoodoutpositive
3、lyinsomestages.Themainmotivationofthispaperistheextensionoffuzzylogicalgorithmtoimproveservocontrolperformance in industrial applications. Keywords-Fuzzy-PI； Speed Control； Disturbance；PMSM Introduction High-performanceservosystemforpermanentmagneticsynchronousmotor(PMSM)isessentialinmanyapplication
4、sinthefieldofmechatronicssuchasprecisionengineering,computernumericallycontrolledmachinetoolsandotherapplicationsinavarietyofautomatedindustrialplants.Duetotheuncertainties,whicharecomposedofunpredictableplantparametervariations,loaddisturbances,andnonlineardynamicsoftheplant ,thecontrolperformanceo
5、fPMSMservosystemisinfluencedseriously.Inthissituation,theservodrivemayneedtorespondrelativelyswiftlytocommandchangesandtoofferenoughrobustnessagainsttheuncertainties.Inordertomeetthedevelopmentrequirementsofhighspeed andhighprecisionforlinearmotor,itisthusdesiredtohaveanintelligentcontrollerthatcano
6、wnhigheranti-disturbanceperformanceaccordingtothedisturbancesanduncertainties in operating condition. Up tonow, a largenumberofcontroltechniques(fuzzy,PI,PID,etc.)withvaryingcomplexityhavebeenproposed.Fuzzycontrolwasfirstintroducedandimplementedintheearly1970inanattempttodesigncontrollersforsystemst
7、hatarestructurallydifficulttomodelduetonaturallyexistingnonlinearitiesandothermodelingcomplexities.Santetal.presentthevectorcontrolofPMSMwithhybridfuzzyPIspeedcontrollerwithswitchingfunctionscalculatedbasedontheweights.Yen-ShinLaietal.presentanewhybridPI-typefuzzycontrollerfordirecttorquecontrolindu
8、ctionmotordriveswithfasttrackingcapability,lesssteadystateerror,androbusttoloaddisturbance.Insummary,fuzzylogiccontrolappearsveryusefulwhentheprocessesaretoocomplextoanalyzebyconventionalquantitativetechniques.Itseemscleartoeveryonethatspeedcontroltechniqueshaveallowedtoexecuteincreasinglymorecomple
9、xtasksinservosystemfield. Theperformanceofthefuzzy-PIcontrollersalsodependsonthechoiceofasuitableoptimizationalgorithm.Inthispaper,anadaptivespeedcontrollerisproposedtominimizeor 2 eliminatethespeedtrackingerror.Thedesignedhybridfuzzy-PIcontrollerimprovessystemperformanceinthetransientandsteadystate
10、.Thispaperisorganizedasfollows.Insection2,thevectorcontrolanddisturbanceeffectsforPMSMaredescribedindetail.Theadaptivefuzzy-PIcontrollerisexplainedinsection3,whilstexperimentalresultsarepresentedinsection4andconclusionsaredrawninthefinalsection. Pmsmvector control InthePMSM,excitationfluxisset-upbym
11、agnets；subsequentlymagnetizingcurrentisnotneededfromthesupply.Thiseasilyenablestheapplicationofthefluxorientationmechanismbyforcingthemagnetizingcurrentcomponentofthestatorcurrentvectortobezero.Asaresult,theelectromagnetictorquewillbedirectlyproportionaltothetorquecurrentcomponentofthestatorcurrentv
12、ector,hencebetterdynamicperformanceisobtainedbycontrollingtheelectromagnetictorqueseparately.AsystemconfigurationofavectorcontrolPMSMservosystemisshowninFig.1.Inthevectorcontrolscheme,torquecontrolcanbecarriedoutbysuitableregulationofthestatorcurrentvector；thisimpliesthataccuratespeedcontroldependso
13、n the regulated current vector. r f iqr idr=0 iqf idf ia ib Figure 1. The systemconfiguration ofavectorcontrol PMSM SpeedcontrolsystemofPMSMisalsomulti-variable,nonlinear,strong-coupledsystem,andthedisturbancesmainlyincludetheloadinertiaandloadtorque.Intherunningofservosystem,systeminertiamaychange.
14、Whenthesysteminertiaincreases,theresponseofservosystemwillslowdown,whichislikelytocausesysteminstabilityandresultinclimb.Onthecontrary,whenthesysteminertiadecreases,dynamicresponsewillspeedupwithspeedovershootaswellasturbulence.Meanwhile,themainroleofservosystemistodrivetheloadoperation,butinmanyind
15、ustries,theloadcarriedbyservosystem isnotconstant.Changesintheloadtorquewillhavesignificantimpactonservocontrolperformance:intherunningofservosystem,thesuddenincreaseorreduceofloadtorquewouldresultinfluctuationsinservospeed control, affecting the accuracy ofpositioning and control performance. Desig
16、n of speed controller Inthispaper,weareproposingaspeedcontrolschemebasedonfuzzylogictoimprovethecontrolperformanceforPMSM.Speedcontroller can be implemented using severalapproaches,such as PI, fuzzy, etc.However, when implementing a speed controller thefollowingconditionsshouldbeconsidered: Simplici
17、ty:Thespeedcontrollawmustbesimpleandeasytocomputeinordertoenablefastservoadaptation. PI-typecontrol:Inordertoachieveanullsteadystateerror,aPItypespeedcontrollaw 3 i (i) should be selected and implemented. Implementationrequirementsshouldnotincludesignificantchangestotheoriginalcontrol system. Giveno
18、urobjectiveandsystem requirements,twocontrolalgorithms,PIandfuzzylogic,arechosen.ThechoiceforPIcontrollerisduetoitsgoodperformancewhenappliedinpracticalsituations,andthepreferenceforfuzzycontrollerisduetonorequirementoftherigorous mathematical systemmodel. Fuzzy ControlArchitecture Fuzzylogicwasconc
19、eivedtoapplyamorehuman-likewayofthinkingincomputerprogramming.Itisidealforcontrollingnonlinearsystemsandmodelcomplexsystemswhereambiguityiscommon.Itisalsopotentiallyveryrobust,maintaininggoodclosed-loopsystemperformanceoverawiderangeofoperatingconditions.Inoursystem,speedcontrollerinputvariablesaret
20、hespeederroreandchangeofthespeederrorde: e(k)=r(k)f(k) (1) de(k)=e(k)e(k1) (2) Wherer isthespeedcommandandf istheactualspeed. Fuzzy-PI FromtheconventionalPIcontrolalgorithm,wecanobtainthefollowingdiscreteequations: iq(k)= iq1(k)+iq(k) (3) iq(k)=kpde(k)+kie(k) (4) Ifeanddearefuzzyvariables,(3)and(4)b
21、ecomeafuzzycontrolalgorithm.Then,thecentreofareamethodisselectedfordefuzzifytheoutputfuzzysetinferredbythecontroller: iq n = i=1 i q n i=1 i (5) Wherei isthe membership function,which takesvaluesintheinterval0,1. Knowledge Base Theknowledgebaseoffuzzylogiccontrolleriscomposedoftwocomponents,namely,adatabaseandafuzzycontrolrulebase.Thewell-knownPI-likefuzzyrulebaseisusedinthispaper(Table1).ThesurfaceofrulebaseisshowninFig.2.Itallowsfastworkingconvergencewithoutsignificantoscillationsand prevents overshoots and undershoots. TABLE1 FUZZYRULEBASE
