基于模糊逻辑控制的反应釜温度控制系统外文文献翻译英译汉.docx

基于模糊逻辑控制的反应釜温度控制系统外文文献翻译英译汉.docx

《基于模糊逻辑控制的反应釜温度控制系统外文文献翻译英译汉.docx》由会员分享，可在线阅读，更多相关《基于模糊逻辑控制的反应釜温度控制系统外文文献翻译英译汉.docx（14页珍藏版）》请在冰点文库上搜索。

基于模糊逻辑控制的反应釜温度控制系统外文文献翻译英译汉

外文文献

FuzzyLogicControlSystemforCSTRTemperatureControl

MOLOYDUTTA,VAIBHAVBAPAT,SCACHINSHELAKE,TUSHARACHYUT&PROF.A.D.SONAR

ABSTRACT

Closedloopcontrolsystemincorporatingfuzzylogichasbeendevelopedforaclassofindustrialtemperaturecontrolproblem.Auniquefuzzylogiccontroller（FLC）structurewithanefficientrealizationandasmallrulebasethatcanbeeasilyimplementedinexistingindustrialcontrollerswasproposed.Itwasdemonstratedinbothsoftwaresimulationandhardwaretestinanindustrialsettingthatthefuzzylogiccontroller（FLC）ismuchmorecapablethanthecurrenttemperaturecontroller.Thisincludescompensatingforthermomasschangesinthesystem,dealingwithunknownandvariabledelays,operatingatverydifferenttemperaturesetpointswithoutreturningetc.Itisachievedbyimplementing,inFLC,aclassicalcontrolstrategyandanadaptationmechanismtocompensateforthedynamicchangesinthesystem.TheproposedFLCwasappliedtotemperaturecontrolofcontinuouslystirredtankreactor（CSTR）andsignificantimprovementsinthesystemperformanceareobserved.

INTRODUCTION

Whilemoderncontroltheoryhasmademodestinroadintopractice,fuzzylogiccontrolhasbeenrapidlygainingpopularityamongpracticingengineers.Thisincreasedpopularitycanbeattributedtothefactthatfuzzylogiccontrolprovidesapowerfulvehiclethatallowsengineerstoincorporatehumanreasoninginthecontrolalgorithm.Asopposedtomoderncontroltheory,fuzzylogicdesignisnotbasedonthemathematicalmodeloftheprocess.

Thecontrollerdesignedusingfuzzylogicimplementshumanreasoningthathasbeenprogrammedintofuzzylogiclanguage（membershipfunctions,ruleandtheruleinterpretation）.

Itisinterestingtonotethatthesuccessoffuzzylogiccontrolislargelyduetoawarenesstoitsmanyindustrialapplications.Industrialinterestsinfuzzylogiccontrolasevidencedbythemanypublicationsonthesubjectinthecontrolliteraturehavecreatedawarenessofitsincreasingimportancebytheacademiccommunity.Theresearchresultsoverthelastfewyearshavebeenreportedin[2-4].

Inthispaper,weconcentrateonfuzzylogiccontrolasanalternativecontrolstrategytothecurrentproportion-integral-derivative（PID）methodusedwidelyinindustry.ConsideratypicaltemperaturecontrolapplicationshowninFigure1:

Figure1:

AtypicalTemperatureControl

ThetemperatureismeasuredbyasuitablesensorsuchasThermocouples,Resistancetemperaturedetector,Thermistors,etcandconvertedtoasignalacceptabletothecontroller.Thecontrollercomparesthetemperaturesignaltothedesiredsetpointtemperatureandactuatesthecontrolelement.Thecontrolelementaltersthemanipulatedvariabletochangethequantityofheatbeingaddedtoortakenfromtheprocess.Theobjectiveofthecontrolleristoregulatethetemperatureascloseaspossibletothesetpoint.

PROBLEMUNDERSTUDY

Currently,theclassicalPID（proportional,integral,derivative）controliswidelyusedwithitsgainsmanuallytuned,basedonthethermalmassandthetemperaturesetpoint.EquipmentwithlargethermalcapacitiesrequiredifferentPIDgainsthanequipmentwithsmallthermalcapacities.

Inaddition,equipmentoperationoverwiderangesoftemperature（140to500degrees）,forexample,requiresdifferentgainsatthelowerandhigherendofthetemperaturerangetoavoidovershootsandoscillations.Thisisnecessarysinceevenbrieftemperatureovershootsinitiatenuisancealarmsandcostlyshutdownstotheprocessbeingcontrolled.

Generally,tuningthePIDconstantsforalargetemperaturecontrolprocessiscostlyandtime-consuming.ThetaskisfurthercomplicatedwhenincorrectPIDconstantsaresometimesenteredduetolackofunderstandingoftemperaturecontrolprocess[1].

Thedifficultyindealingwithsuchproblemsiscompoundedwithvariabletimedelaysexistinginmanysuchsystems.Variationsinmanufacturing,newproductdevelopmentandphysicalconstraintsplacetheResistanceTemperatureDetector（RTD）temperaturesensoratdifferentlocations,includingvariabletimedelay（deadtime）inthesystem.

ItisalsowellknownthatPIDcontrollersexhibitpoorperformancewhenappliedtosystemscontainingunknownnonlinearitysuchasdeadzones,saturationandhysteresis.

Itisfurtherunderstoodthatmanytemperaturecontrolprocessarenonlinear.Equalincrementsofheatinput,forexample,donotnecessarilyproduceequalincrementsintemperatureriseinmanyprocesses,atypicalphenomenonofnonlinearsystems.

FUZZYLOGICCONTROL

Fuzzylogiccontrolisanappealingalternativetoconventionalcontrolmethodswhensystemsfollowsomegeneraloperatingcharacteristicsanddetailedprocessunderstandingisunknownortraditionalsystemmodelbecomeoverlycomplex[1,a].Themainfeatureoffuzzycontrolisthecapabilitytoqualitativelycapturetheattributesofacontrolsystembasedonobservablephenomenon[a,b].

FuzzyLogicControlDesign

TheFLCdevelopedhereisatwo-inputandsingle-outputcontroller.Theinputsarethedeviationfromsetpointerror,e（k）anderrorrate,∆e（k）.TheoperationalstructureofthefuzzycontrollerisshowninFigure2:

Figure2:

StructureofFuzzyController

Fuzzification

Fuzzificationinvolvesmappingthefuzzyvariablesofintereststo“crisp”numbersusedbythecontrolsystem.Fuzzificationtranslatesanumbericvaluefortheerror,e（k）,orerrorrate,∆e（k）,intoalinguisticvaluesuchaspositivelargewithamembershipgrade.

TheFLCmembershipfunctionsaredefinedovertherangeofinputandoutputvariablevaluesandlinguisticallydescribesthevariable’suniverseofdiscourseasshowninFigures3、4、5.

Figure3:

MembershipFunctionforError（e）

Figure4:

MembershipFunctionforChangeinError（∆e）

Figure5:

ChangeinOutput（inwant）

TABLE1

FLCCONTROLRULES

e（k）

∆e（k）

NB

NM

NS

ZO

PS

PM

PB

NB

NB

NS

ZO

PB

PB

PB

PB

NM

NB

NS

PB

PB

PB

PB

PB

NS

NB

NS

PB

PB

PB

PB

PB

ZO

NM

NS

PB

PB

PB

PB

PB

PS

NM

ZO

PB

PB

PB

PB

PB

PM

NS

ZO

PB

PB

PB

PB

PB

PB

NS

ZO

PB

PB

PB

PB

PB

Herethetemperaturerangeisfrom0~100℃.Thevalueofmembershipfunctionoferrorvariesfrom-5to75℃andfortheerrorchangeis-5to0℃.

Thetriangularinputmembershipfunctionsforthelinguisticlabelszero,small,mediumandlarge.Theleftandrighthalfofthetriangularforeachlinguisticlabelissochosenthatmembershipoverlapwithadjacentmembershipfunctions.

Theoutputmembershipfunctionsforthelabelsarezero,small,mediumandlarge.Boththeinputandoutputvariablesmembershipfunctionsaresymmetricwithrespecttotheorigin.Selectionofthenumberofmembershipfunctionsandtheirinitialvaluesarebasedonprocessknowledgeandintuition.Themainideaistodefinepartitionofoperatingregionsthatwillrepresenttheprocessvariables.

Rulesdevelopment

Rulesdevelopmentstrategyforsystemswithtimedelayistoregulatetheoverallloopgaintoachievethedesiredstepresponse.TheoutputoftheFLCisbasedonthecurrentinpute（k）and∆e（k）,andwithoutanyknowledgeofthepreviousinputandoutputdata.TherulesdevelopedinthispaperforCSTRareabletocompensateforvaryingtimedelaysonlinebytuningtheFLCoutputmembershipfunctionsbasedonsystemperformance.TheTable1showshowrulesarerepresentedforCSTR[8].

Defuzzification

Defuzzificationtakesthefuzzyoutputoftherulesandgeneratesa“crisp”numbericvalueuseascontrolinputtoplant.

Tuningofmembershipfunction

Themembershipfunctionssubjecttothestabilitycriteriabasedonobservationsofsystemperformancesuchasrisetime,overshoot,steadystateerror.Accordingtotheresolutionneeded,numberofmembershipfunctionincreases.Thecenterandslopesoftheinputmembershipfunctionsineachregionisadjustedsothatthecorrespondingruleprovidesanappropriatecontrolaction.Incasewhentwoormorerulesarefiredatthesametime,thedominantruleistunedfirst.Onceinputmembershipruletuningiscompleted,fine-tuningofoutput,membershipfunctionisperformed.

APPLICATION

CSTRtemperaturecontrolhardwaresetup

AloseloopdiagramoftheprocessisshowninFigure6:

Figure6:

Closed-loopTemperatureControlSystem

Inthispaper,theapplicationoffuzzylogicistocontrolthetemperatureofwater.ForsensingthetemperatureRTD（ResistanceTemperatureDetector）isusedassensor.Therearemanyvariationsinthedynamicsofthesystem.Thethermocapacityisproportionaltothesizeofthetank.ThetimedelayinthesystemisquitesensitivetotheplacementoftheRTD.TheRTDsensesthetemperatureofwaterandgivethesignaltotheFLC（FuzzyLogicController）anditcalculatesthe“crisp”value.Dependinguponon“crisp”value,firingangleofSCR（SiliconControlledRectifier）ischangingandeventuallycontrolthepowersuppliedtotheheaterthroughinterfacingcard.

TESTRESULTS

Intemperaturecontrolapplication,itisimportanttopreventovershoots,whichseriouslyaffectthesystemperformance.Itisalsodesirabletohaveasmoothcontrolsignalthatdoesnotrequireexcessiveonandoffactionsintheheater.TheresultsareshownintheFigure7.Ineachcase,theFLCwasabletosuccessfullymeetalldesignspecificationswithoutoperatortuning.

Figure7:

ProcessResponse

CONCLUSION

Fuzzyprovidesaremarkablysimplewaytodrawdefiniteconclusionsfromvague,ambiguous,impreciseinformation.Inasense,fuzzylogicresembleshumandecisionmakingwithitsabilitytoworkfromapproximatedataandfindprecisesolution.Theresultsshowsignificantimprovementinmaintainingperformanc