電機(jī)及其傳動(dòng)系統(tǒng) 原理、控制、建模和仿真課件：Simulation-Based Design of Electric Drive Systems

Simulation-BasedDesignofElectricDriveSystemsDesignofdrivesystemsSelectionofcomponentsizesFilterandenergystorageelementsinthedrivecircuit;SwitchingfrequencyControllertuningControllertuningforimproved/optimaldynamicresponse;RobustoptimalcontrolDesigncompromiseCostvs.performanceSensitivityvs.performanceDesignmethodologies(1)Simplified-modelbasedAsimplified,low-ordermathematicalmodelisusedasthebasis;Pros:Simple,inexpensive,easytouseCons:Lessaccuracy,andun-modeleddynamicsparticularlyinhighfrequencies;Thefinaldesignmaynotperformasintendedduetotheabove.Designmethodologies(2)Simulation-modelbasedAdetailedcomputersimulationmodelisusedasthebasis;Pros:Noneedtodevelopamathematicalmodel;Higheraccuracy;Inclusionofhigher-orderdynamics;Cons:Lessintuitive;potentiallytimeconsuming;Simulation-baseddesignapproaches(1)TrialanderrorEngineeringjudgmentisusedtoconduct(alargenumberof)simulation-basedexperimentstodetermineasuitabledesign;Pros:Humanexpertiseisfullydeployedinsteeringthesequenceofsimulationexperiments;Cons:Timeconsuming;Mayinhibitexhaustivesearch.Simulation-baseddesignapproaches(2)Simulation-basedoptimaldesignAnoptimizationalgorithmiscoupledwithahigh-fidelitysimulationmodeltoconductastrategicsequenceofsimulationexperimentstoyieldanoptimaldesign;Pros:Fullyautomatedsearch;Rapidconvergencetotheoptimaldesign;Cons:Requiresexpertiseindefiningaproperobjectivefunction.Simulation-basedoptimaldesignnonlinearoptimizationalgorithmparametersetxsimulationmodelofthedrivesystemobjectivefunctionf(x)ncontrollerparameters,componentsizes,switchingfrequency,etc.figureofmeritforthecurrentlyusedparametersetDevelopmentofanobjectivefunctionTheobjectivefunctionisamathematicalrepresentationofthedesigngoals;Mustbeevaluatedthroughsimulation;Neednotbeexplicitlyavailableasafunctionoftheoptimizationparameters;Example:PropertiesoftheoptimizationalgorithmSinceexplicitformulationisnotavailable,non-gradientoptimizationalgorithmsarepreferred;Globalvs.localoptimizationalgorithms:Global:Searchesfortheglobalsolution;theonewiththebestoutcome;Local:Searchesforalocalsolution.Examplecase1IndirectvectorcontroldrivesystemoptimizationCasespecifics(1)Inductionmachineparameters2300V,500hp,4-pole,60Hz,rs=0.262W,rr=0.187W,Xls

=1.206W,Xlr

=1.206W,XM=54.02WCasespecifics(2)Circuitdiagramdiode

bridgeLdc+-vdciaCdcCdcVSCinduction

machine+rotor

speed(elec.

rad/s)+iqs-refids-reflr-ref

=ldr-refTe-refqe

we

wr

wref+PI

speed

controller-qdoàabcPWMvabc-refgate

pulsesPI

controller

+

decouplingPI

controller

+

decouplingvqs-refvds-refindirectvector

control+-iq-id+Designobjectives(1)Toensurethatthespeedfollowsthereferenceascloselyaspossible;Tominimizethetorqueripple;Toreducetherippleonthedcvoltageacrossthedclinkcapacitors;Todotheabovewithsmallenergystoringelements.Designobjectives(2)ObjectivefunctionSub-objectivesResults(1)BeforeoptimizationAfteroptimizationResults(2)Thinline:beforeoptimizationThinkline:AfteroptimizationResults(3)Parameters

KqTqKdTdKwTwLdcCdcInitial300.5300.52000.51mH500mFOptimized6.931.28264.70.68280.81.960.5mH4269mFObjectives

f1f2f3f4fInitial0.000430.00910.041550.6189.72Optimized0.0000380.00590.000481.10523.25Examplecase2Multi-objectiveoptimizationRationale Designoftenpursuesmorethanasingleobjective;Designobjectivesarecombinedinanaggregateobjectivefunction;Propercompromisemustbemadebetweenthecompetingobjectives.Casespecifics(1)Ascalarinductionmachinedrivewiththefollowingobjectives:Toensurerapidandsmoothtransientresponsetotorquecommands;Toobtainsmallrippleonthedcvoltageandcurrentinsteadystate.Thetwoobjectivespresentaconflictingcase:Largerenergystorageelementsimprovetheripple;Largerenergystorageelementsmakethedynamicresponseslow.Casespecifics(2)ObjectivefunctionCombinationofsteadystate(ripple)anddynamic(controlresponse)Bychangingkover[0,1],thefocusoftheoptimizationalgorithmshiftsfromfss(k≈1)toftr

(k≈0);Itisaminimizationproblem.Results(1)TheParetofrontierpoorftrandgoodfsscompromisegoodftrandpoorfssResults(2)PointAPointCPointBExamplecase3 MultipleoptimalsolutionsRationaleObjectivefunctionsmayhavemorethanlocallyoptimalsolution;Globaloptimumhasthebestquality,but:Itmaybeimpracticaltoimplement,e.g.componentsizestoolargeItmaybeexcessivelysensitive:anysmalldeviationintheparameterscausesdeteriorationoftheresultsItisthereforeimportanttoobtainasmanyoptimalsolutionsaspossible.CasespecificsDesignobjectivesToensurethatthespeedfollowsthereferenceascloselyaspossible;To