abstract
Abstract
Vectorcontrolstrategyisacontrolstrategyofpermanentmagnetsynchronous
motor(PMSM),whichiswidelystudiedandappliedathomeandabroad.Inthispaper,
thedeadbeatcontrolofPMSMbasedonneuralslidingmodeobserverisdesigned,the
deadbeatpredictivecontrolalgorithmisusedtodesignthecontroller,andthesliding
modecontrolcombinedwithneuralnetworkalgorithmisusedtodesigntheobserver.it
effectivelysolvestheproblemsofhighovershoot,slowresponseandstrongbuffetingof
traditionalPMSMvectorcontrolstrategy.Inthispaper,theproposedalgorithmis
simulated,andtheeffectivenessofthemethodisproved.
Firstofall,byanalyzingthecurrentdynamicequationofthree-phasepermanent
magnetsynchronousmotor,thecurrentpredictionequationisconstructed,andthe
deadbeatcontrollerisdesigned.ThedynamicandstaticperformanceofPMSMunder
deadbeatcontrolisanalyzedandstudiedbysimulation.Thesimulationresultsshowthat
thedeadbeatcontrolcanimprovetheresponsespeedofthemotorandreducethe
overshoot.
Secondly,thispaperproposesaslidingmodeobservercombinedwithradialbasis
functionneuralnetworkalgorithm.Onthebasisofthetraditionalslidingmodeobserver,
theradialbasisfunctionneuralnetworkisusedtocompensatetheoutputofthePI
controller.Atthesametime,thesaturationfunctionsat(s)isusedinsteadofthe
switchingfunctionsign(s),andtheexponentialreachinglawisadoptedtoimprovethe
observationaccuracy,andthenimprovethedynamicandstaticperformanceofthe
motor.Thesimulationresultsshowthattheslidingmodeobserverdesignedinthispaper
canachievefastandaccuratetracking.
Keywords:Permanentmagnetsynchronousmotor;vectorcontrol;deadbeat;sliding
modeobserver;neuralnetwork;exponentialapproachrate