机械类文献翻译PLC变频调速的网络反馈系统的实现.docx
《机械类文献翻译PLC变频调速的网络反馈系统的实现.docx》由会员分享,可在线阅读,更多相关《机械类文献翻译PLC变频调速的网络反馈系统的实现.docx(16页珍藏版)》请在冰豆网上搜索。
机械类文献翻译PLC变频调速的网络反馈系统的实现
英文原文:
RealizationofNeuralNetworkInverseSystemwithPLCinVariableFrequencySpeed-RegulatingSystem
Abstract.Thevariablefrequencyspeed-regulatingsystemwhichconsistsofaninductionmotorandageneralinverter,andcontrolledbyPLCiswidelyusedinindustrialfield..However,forthemultivariable,nonlinearandstronglycoupledinductionmotor,thecontrolperformanceisnotgoodenoughtomeettheneedsofspeed-regulating.Themathematicmodelofthevariablefrequencyspeed-regulatingsysteminvectorcontrolmodeispresentedanditsreversibilityhasbeenproved.Byconstructinganeuralnetworkinversesystemandcombiningitwiththevariablefrequencyspeed-regulatingsystem,apseudo-linearsystemiscompleted,andthenalinearclose-loopisdesignedtogethighperformance.UsingPLC,aneuralnetworkinversesystemcanberealizedinsystem.Theresultsofexperimentshaveshownthattheperformancesofvariablefrequencyspeed-regulatingsystemcanbeimprovedgreatlyandthepracticabilityofneuralnetworkinversecontrolwastestified.
1.Introduction
Inrecentyears,withpowerelectronictechnology,microelectronictechnologyandmoderncontroltheoryinfiltratingintoACelectricdrivingsystem,invertershavebeenwidelyusedinspeed-regulatingofACmotor.Thevariablefrequencyspeed-regulatingsystemwhichconsistsofaninductionmotorandageneralinverterisusedtotaketheplaceofDCspeed-regulatingsystem.Becauseofterribleenvironmentandseveredisturbanceinindustrialfield,thechoiceofcontrollerisanimportantproblem.Inreference[1][2][3],Neuralnetworkinversecontrolwasrealizedbyusingindustrialcontrolcomputerandseveraldataacquisitioncards.Theadvantagesofindustrialcontrolcomputerarehighcomputationspeed,greatmemorycapacityandgoodcompatibilitywithothersoftwareetc.Butindustrialcontrolcomputeralsohassomedisadvantagesinindustrialapplicationsuchasinstabilityandfallibilityandworsecommunicationability.PLCcontrolsystemisspecialdesignedforindustrialenvironmentapplication,anditsstabilityandreliabilityaregood.PLCcontrolsystemcanbeeasilyintegratedintofieldbuscontrolsystemwiththehighabilityofcommunicationconfiguration,soitiswildlyusedinrecentyears,anddeeplywelcomed.Sincethesystemcomposedofnormalinverterandinductionmotorisacomplicatednonlinearsystem,traditionalPIDcontrolstrategycouldnotmeettherequirementforfurthercontrol.Therefore,howtoenhancecontrolperformanceofthissystemisveryurgent.
Theneuralnetworkinversesystem[4][5]isanovelcontrolmethodinrecentyears.Thebasicideaisthat:
foragivensystem,aninversesystemoftheoriginalsystemiscreatedbyadynamicneuralnetwork,andthecombinationsystemofinverseandobjectistransformedintoakindofdecouplingstandardizedsystemwithlinearrelationship.Subsequently,alinearclose-loopregulatorcanbedesignedtoachievehighcontrolperformance.Theadvantageofthismethodiseasilytoberealizedinengineering.Thelinearizationanddecouplingcontrolofnormal
systemcanrealizeusingthismethod.
CombiningtheneuralnetworkinverseintoPLCcaneasilymakeuptheinsufficiencyofsolvingtheproblemsofnonlinearandcouplinginPLCcontrolsystem.Thiscombinationcanpromotetheapplicationofneuralnetworkinverseintopracticetoachieveitsfulleconomic.
Inthispaper,firstlytheneuralnetworkinversesystemmethodisintroduced,andmathematicmodelofthevariablefrequencyspeed-regulatingsysteminvectorcontrolmodeispresented.Thenareversibleanalysisofthesystemisperformed,andthemethodsandstepsaregiveninconstructingNN-inversesystemwithPLCcontrolsystem.Finally,themethodisverifiedin
traditionalPIcontrolandNN-inversecontrol.
2.NeuralNetworkInverseSystemControlMethod
Thebasicideaofinversecontrolmethod[6]isthat:
foragivensystem,anα-thintegralinversesystemoftheoriginalsystemiscreatedbyfeedbackmethod,andcombiningtheinversesystemwithoriginalsystem,akindofdecouplingstandardizedsystemwithlinearrelationshipisobtained,whichisnamedasapseudolinearsystemasshowninFig.1.Subsequently,alinearclose-loopregulatorwillbedesignedtoachievehighcontrolperformance.
Inversesystemcontrolmethodwiththefeaturesofdirect,simpleandeasytounderstanddoesnotlikedifferentialgeometrymethod[7],whichisdiscussestheproblemsin"geometrydomain".Themainproblemistheacquisitionoftheinversemodelintheapplications.Sincenon-linearsystemisacomplexsystem,anddesiredstrictinverseisverydifficultto
obtain,evenimpossible.Theengineeringapplicationofinversesystemcontroldon’tmeettheexpectations.Asneuralnetworkhasnon-linearapproximateability,especiallyfornonlinear
thepowerfultooltosolvetheproblem.
a−thNNinversesystemintegratedinversesystemwithnon-linearabilityoftheneuralnetworkcanavoidthetroublesofinversesystemmethod.Thenitispossibletoapplyinversecontrolmethodtoacomplicatednon-linearsystem.a−thNNinversesystemmethodneedslesssysteminformationsuchastherelativeorderofsystem,anditiseasytoobtaintheinversemodelbyneuralnetworktraining.CascadingtheNNinversesystemwiththeoriginalsystem,apseudo-linearsystemiscompleted.Subsequently,alinearclose-loopregulatorwillbedesigned.
3.MathematicModelofInductionMotorVariableFrequency
Speed-RegulatingSystemandItsReversibility
Inductionmotorvariablefrequencyspeed-regulatingsystemsuppliedbytheinverteroftrackingcurrentSPWMcanbeexpressedby5thordernonlinearmodelind-qtwo-phaserotatingcoordinate.Themodelwassimplifiedasa3-ordernonlinearmodel.Ifthedelayofinverterisneglected,
themodelisexpressedasfollows:
(1)
where
denotessynchronousanglefrequency,and
isrotatespeed.
arestator’scurrent,and
arerotor’sfluxlinkagein
(d,q)axis.
isnumberofpoles.
ismutualinductance,and
isrotor’sinductance.Jismomentofinertia.
isrotor’stimeconstant,and
isloadtorque.
Invectormode,then
Substituteditintoformula
(1),then
(2)
Takingreversibilityanalysesofforum
(2),then
Thestatevariablesarechosenasfollows
Inputvariablesare
Takingthederivativeonoutputinformula(4),then
(5)
(6)
ThentheJacobimatrixisRealizationofNeuralNetworkInverseSystemwithPLC
(7)
(8)
As
so
andsystemisreversible.Relative-orderofsystemis
Whentheinverterisrunninginvectormode,thevariabilityoffluxlinkagecanbeneglected(consideringthefluxlinkagetobeinvariablenessandequaltotherating).Theoriginalsystemwassimplifiedasaninputandanoutputsystemconcludedbyforum
(2).
Accordingtoimplicitfunctionontologytheorem,inversesystemofformula(3)
canbeexpressedas
(9)
Whentheinversesystemisconnectedtotheoriginalsysteminseries,thepseudolinearcompoundsystemcanbebuiltasthetypeof
4.RealizationStepsofNeuralNetworkInverseSystem
4.1AcquisitionoftheInputandOutputTrainingSamples
Trainingsamplesareextremelyimportantinthereconstructionofneuralnetworkinversesystem.Itisnotonlyneedtoobtainthedynamicdataoftheoriginalsystem,butalsoneedtoobtainthestaticdate.Referencesignalshouldincludealltheworkregionoforiginalsystem,whichcanbeensuretheapproximateability.Firstlythestepofactuatingsignalisgivencorrespondingevery10HZform0HZto50HZ,andtheresponsesofopenloopareobtain.Secondlyarandomtanglesignalisinput,whichisarandomsignalcascadingonthestepofactuatingsignalevery10seconds,andthecloseloopresponsesisobtained.Basedontheseinputs,1600groups
trainingsamplesaregotten.
4.2TheConstructionofNeuralNetwork
Astaticneuralnetworkandadynamicneuralnetworkcomposedofintegralisusedtoconstructtheinversesystem.Thestructureofstaticneuralnetworkis2neuronsininputlayer,3neuronsinoutputlayer,and12neuronsinhiddenlayer.Theexcitationfunctionofhiddenneuronismonotonicsmoothhyperbolictangentfunction.Theoutputlayeriscomposedofneuronwithlinearthresholdexcitationfunction.Thetrainingdatumarethecorrespondingspeedofopen-loop,close-loop,firstorder
derivativeofthesespeed,andsettingreferencespeed.After50timestraining,thetrainingerrorofneuralnetworkachievesto0.001.Theweightandthresholdoftheneuralnetworkaresaved.Theinversemodeloforiginal
systemisobtained.
5.ExperimentsandResults
5.1HardwareoftheSystem
ThehardwareoftheexperimentsystemisshowninFig5.ThehardwaresystemincludesuppercomputerinstalledwithSupervisory&ControlconfigurationsoftwareWinCC6.0[8],andS7-300PLCofSIEMENS,inverter,inductionmotorandphotoelectriccoder.
PLCcontrollerchoosesS7-315-2DP,whichhasaPROFIBUS-DPinterfaceandaMPI
isconnectedwithS7-300byCP5611usingMPIprotocol.
ThetypeofinverterisMMVofSIEMENS.ItcancommunicatewithSIEMENSPLCby
inverterinthissystem.
5.2SoftwareProgram
5.2.1CommunicationIntroduction
MPI(MuPointInterface)isasimpleandinexpensivecommunicationstrategyusinginslowlyandnon-largedatatransformingfield.ThedatatransformingbetweenandPLCisnotlarge,
chosen.
TheMMVinverterisconnectedtothePROFIBUSnetworkasaslavestation,whichismountedwithCB15PROFIBUSmodule.PPO1orPPO3datatypecanbechosen.Itpermitstosendthecontroldatadirectlytotheinverteraddresses,ortousethesystemfunctionblocksof
SFC14/15.
OPCcanefficientlyprovidedataintegrala