会员注册 | 登录 | 微信快捷登录 QQ登录 微博登录 | 帮助中心 人人文库renrendoc.com美如初恋!
站内搜索 百度文库

热门搜索: 直缝焊接机 矿井提升机 循环球式转向器图纸 机器人手爪发展史 管道机器人dwg 动平衡试验台设计

外文翻译--内嵌于可编程控制器的先进控制算法 英文文版.pdf外文翻译--内嵌于可编程控制器的先进控制算法 英文文版.pdf -- 5 元

宽屏显示 收藏 分享

资源预览需要最新版本的Flash Player支持。
您尚未安装或版本过低,建议您

ControlEngineeringPracticeembeddedVrancLjubljana,bNovaGoricaPolytechnic,NovaGorica,Sloveniaidentificationstepstoprovidereliableoperation.Thecontrollermonitorsandevaluatesthecontrolperformanceoftheclosedloopsystem.ThecontrollerwasimplementedonaprogrammablelogiccontrollerPLC.Theperformanceisillustratedonafieldtestinindustrialapplications,assummarisedbelowARTICLEINPRESSwww.elsevier.com/locate/conengpracC3Correspondingauthor.Tel.386147739941.Becauseofthediversityofreallifeproblems,asinglenonlinearcontrolmethodhasarelativelynarrow09670661/seefrontmatterr2005ElsevierLtd.Allrightsreserved.doi10.1016/j.conengprac.2005.05.006fax38614257009.Emailaddresssamo.gerksicijs.siS.Gerksˇicˇ.applicationforcontrolofpressureonahydraulicvalve.r2005ElsevierLtd.Allrightsreserved.KeywordsControlengineeringFuzzymodellingIndustrialcontrolModelbasedcontrolNonlinearcontrolProgrammablelogiccontrollersSelftuningregulators1.IntroductionModerncontroltheoryoffersmanycontrolmethodstoachievemoreefficientcontrolofnonlinearprocessesthanprovidedbyconventionallinearmethods,takingadvantageofmoreaccurateprocessmodelsBequette,1991HensonSeborg,1997MurraySmithJohansen,1997.SurveysTakatsu,Itoh,Araki,1998Seborg,1999indicatethatwhilethereisaconsiderableandgrowingmarketforadvancedcontrollers,relativelyfewvendorsofferturnkeyproducts.Excellentresultsofadvancedcontrolconcepts,basedonfuzzyparameterschedulingTan,Hang,Chai,1997Babusˇka,Oosterhoff,Oudshoorn,Bruijn,2002,multiplemodelcontrolDoughertyCooper,2003Gundala,Hoo,Piovoso,2000,andadaptivecontrolHensonSeborg,1994Ha¨gglundA˚strom,2000,havebeenreportedintheliterature.However,thereareseveralrestrictionsforapplyingthesemethodsdINEAd.o.o.,Ljubljana,SloveniaeComputerTechnologyInstitute,Athens,GreecefUniversityofChemicalTechnologyandMetallurgySofia,Sofia,BulgariaReceived23April2004accepted15May2005AbstractThispaperpresentsaninnovativeselftuningnonlinearcontrollerASPECTadvancedcontrolalgorithmsforprogrammablelogiccontrollers.Itisintendedforthecontrolofhighlynonlinearprocesseswhosepropertieschangeradicallyoveritsrangeofoperation,andincludesthreeadvancedcontrolalgorithms.Itisdesignedusingtheconceptsofagentbasedsystems,appliedwiththeaimofautomatingsomeoftheconfigurationtasks.Theprocessisrepresentedbyasetofloworderlocallinearmodelswhoseparametersareidentifiedusinganonlinelearningprocedure.ThisprocedurecombinesmodelidentificationwithpreandpostcUniversityofLjubljana,FacultyofElectricalEngineering,Ljubljana,SloveniaAdvancedcontrolalgorithmslogiccontrollerSamoGerksˇicˇa,C3,GregorDolanca,DamirSasˇoBlazˇicˇc,IgorSˇkrjancc,ZoranMarinsRobertKinge,MinchoHadjiskiaJozˇefStefanInstitute,inaprogrammableˇic´a,JusˇKocijana,b,StankoStrmcˇnika,ˇekd,MihaBozˇicˇekd,AnnaStathakie,f,KostaBoshnakovfSlovenia–friendmaticindustfromling,procedcontrollermonitorstheresultingcontrolperformanceARTICLEINPRESSanonlinearprocessmodel.Themodelisobtainedoperatingprocesssignalsbyexperimentalmodelusinganovelonlinelearningprocedure.ThisThefromforimplementationonPLCoropencontrollerrialhardwareplatforms.controllerparametersareautomaticallytunedfeatuadaptedssioningofthecontrollerissimplifiedbyautoexperimentationandtuning.AdistinguishingreofthecontrolleristhatthealgorithmsaremetecommiTheASPECTcontrollerisanefficientanduserlyengineeringtoolforimplementationofpararschedulingcontrolintheprocessindustry.Theused,thesensorreadings,specifichardwareplatformsareetc.isdemandedtofieldofapplication.Therefore,moreflexiblemethodsoratoolboxofmethodsarerequiredinindustry.2.Newmethodsareusuallynotavailableinareadytouseindustrialform.Customdesignrequiresconsiderableeffort,timeandmoney.3.Thehardwarerequirementsarerelativelyhigh,duetothecomplexityofimplementationandcomputationaldemands.4.ThecomplexityoftuningBabusˇkaetal.,2002andmaintenancemakesthemethodsunattractivetononspecialisedengineers.5.Thereliabilityofnonlinearmodellingisofteninquestion.6.ManynonlinearprocessescanbecontrolledusingthewellknownandindustriallyprovenPIDcontroller.Aconsiderabledirectperformanceincreasefinancialgainisdemandedwhenreplacingaconventionalcontrolsystemwithanadvancedone.Themaintenancecostsofaninadequateconventionalcontrolsolutionmaybelessobvious.TheaimofthisworkistodesignanadvancedcontrollerthataddressessomeoftheaforementionedproblemsbyusingtheconceptsofagentbasedsystemsABSWooldridgeJennings,1995.Themainpurposeistosimplifycontrollerconfigurationbypartialautomationofthecommissioningprocedure,whichistypicallyperformedbythecontrolengineer.ABSsolvedifficultproblemsbyassigningtaskstonetworkedsoftwareagents.Thesoftwareagentsarecharacterisedbypropertiessuchasautonomyoperationwithoutdirectinterventionofhumans,socialabilityinteractionwithotheragents,reactivityperceptionandresponsetotheenvironment,proactivenessgoaldirectedbehaviour,takingtheinitiative,etc.ThisworkdoesnotaddressissuesofABStheory,butrathertheapplicationofthebasicconceptsofABStothefieldofprocesssystemsengineering.Inthiscontext,anumberoflimitshavetobeconsidered.Forexampleinitiativeisrestricted,ahighdegreeofreliabilityandpredictability,insightintotheproblemdomainislimitedS.Gerksˇicˇetal./ControlEngineerin2ureisbasedonmodelidentificationusingtheandreactstodetectedirregularities.ThecontrollercomprisestheruntimemoduleRTMandtheconfigurationtoolCT.TheRTMrunsonaPLC,performingallthemainfunctionalityofrealtimecontrol,onlinelearningandcontrolperformancemonitoring.TheCT,usedonapersonalcomputerPCduringtheinitialconfigurationphase,simplifiestheconfigurationprocedurebyprovidingguidanceanddefaultparametervalues.TheoutlineofthepaperisasfollowsSection2presentsanoverviewoftheRTMstructureanddescribesitsmostimportantmodulesSection3givesabriefdescriptionoftheCTandfinally,Section4describestheapplicationofthecontrollertoapilotplantwhereitisusedforcontrolofthepressuredifferenceonahydraulicvalveinavalvetestapparatus.2.RunTimeModuleTheRTMoftheASPECTcontrollercomprisesasetofmodules,linkedintheformofamultiagentsystem.Fig.1showsanoverviewoftheRTManditsmainmodulesthesignalpreprocessingagentSPA,theonlinelearningagentOLA,themodelinformationagentMIA,thecontrolalgorithmagentCAA,thecontrolperformancemonitorCPM,andtheoperationsupervisorOS.2.1.MultifacetedmodelMFMTheASPECTcontrollerisbasedonthemultifacetedmodelconceptproposedbyStephanopoulus,Henning,andLeone1990andincorporatesseveralmodelformsrequiredbytheCAAandtheOLA.Specifically,theMFMincludesasetoflocalfirstandsecondorderlocallearningapproachMurraySmithJohansen,1997,p.188.Themeasurementdataareprocessedbatchwise.Additionalstepsareperformedbeforeandafteridentificationinordertoimprovethereliabilityofmodelling,comparedtoadaptivemethodsthatuserecursiveidentificationcontinuouslyHa¨gglundA˚strom,2000.Thenonlinearmodelcomprisesasetoflocalloworderlinearmodels,eachofwhichisvalidoveraspecifiedoperatingregion.Theactivelocalmodelsisselectedusingaconfiguredschedulingvariable.Thecontrollerisspecificallydesignedforsingleinput,singleoutputprocessesthatmayincludeameasureddisturbanceusedforfeedforward.Additionally,theapplicationrangeofthecontrollerdependsontheselectedcontrolalgorithm.Amodularstructureofthecontrollerpermitsuseofarangeofcontrolalgorithmsthataremostsuitablefordifferentprocesses.ThegPractice–discretetimelinearmodelswithtimedelayandoffset,ARTICLEINPRESSS.Gerksˇicˇetal./ControlEngineerinwhicharespecifiedbyagivenschedulingvariablesk.Themodelequationoffirstorderlocalmodelsisyðkþ1Þ¼C0a1jyðkÞþb1juðkC0dujÞþc1jvðkC0dvjÞþrj,1whilethemodelequationofsecondordermodelsisyðkþ1Þ¼C0a1jyðkÞC0a2jyðkC01Þþb1juðkC0dujÞþb2juðkC01C0dujÞþc1jvðkC0dvjÞþc2jvðkC01C0dvjÞþrj,ð2Þwherekisthediscretetimeindex,jisthenumberofthelocalmodel,ykistheprocessoutputsignal,ukistheprocessinputsignal,vkistheoptionalmeasureddisturbancesignalMD,duisthedelayinthemodelbranchfromutoy,dvisthedelayinthemodelbranchfromvtoy,andai,j,bi,j,ci,jandrjaretheparametersofthejthlocalmodel.ThesetoflocalmodelscanbeinterpretedasaTakagi–Sugenofuzzymodel,whichinthecaseofasecondordermodelcanbeexpressedintheFig.1.RuntimemodulegPractice–3followingformyðkþ1Þ¼C0Xmj¼1bjðkÞa1jyðkÞC0Xmj¼1bjðkÞa2jyðkC01ÞþXmj¼1bjðkÞb1juðkC0dujÞþXmj¼1bjðkÞb2juðkC01C0dujÞþXmj¼1bjðkÞc1jnðkC0dnjÞþXmj¼1bjðkÞc2jnðkC01C0dnjÞþXmj¼1bjðkÞrj,ð3Þwherebjkisthevalueofthemembershipfunctionofthejthlocalmodelatthecurrentvalueoftheschedulingvariablesk.Normalisedtriangularmembershipfunctionsareused,asillustratedinFig.2.overview.ARTICLEINPRESSTheschedulingvariableskiscalculatedusingcoefficientskr,ky,ku,andkv,usingtheweightedsumsðkÞ¼krrðkÞþkyyðkÞþkuuðkC01ÞþkvvðkÞ.4Thecoefficientsareconfiguredbytheengineeraccordingtothenatureoftheprocessnonlinearity.2.2.OnlineLearningAgentOLATheOLAscansthebufferofrecentrealtimesignals,preparedbytheSPA,andestimatestheparametersofthelocalmodelsthatareexcitedbythesignals.ThemostrecentlyderivedparametersaresubmittedtotheMIAonlywhentheypasstheverificationtestandareprovedtobebetterthantheexistingset.TheOLAisinvokedupondemandfromtheOSorautonomously,whenanintervaloftheprocesssignalswithsufficientexcitationisavailableforprocessing.Itprocessesthesignalsbatchwiseandusingthelocallearningapproach.Anadvantageofthebatchwiseconceptisthatthedecisiononwhethertoadaptthemodelisnotperformedinrealtimebutfollowingadelaythatallowsforinspectionoftheidentificationresultbeforeitisapplied.Thus,bettermeansforcontroloverdataselectionisprovided.Aproblemofdistributionofthecomputationtimerequiredforidentificationappearswithbatchwiseprocessingofdataopposedtotheonlinerecursiveprocessingthatistypicallyusedinadaptivecontrollers.Thisproblemisresolvedusingamultitaskingoperationsystem.SincetheOLAtypicallyrequiresconsiderablyFig.2.FuzzymembershipfunctionsoflocalmodelsintheMFM.S.Gerksˇicˇetal./ControlEngineerin4morecomputationthantherealtimecontrolalgorithm,itrunsinthebackgroundasalowprioritytask.Thefollowingprocedure,illustratedinFig.3,isexecutedwhentheOLAisinvoked.2.2.1.CopysignalbufferThebufferoftherealtimesignalsismaintainedbytheSPA.WhentheOLAisinvoked,therelevantsectionofthebufferiscopiedforfurtherprocessing.2.2.2.ExcitationcheckAquickexcitationcheckisperformedatthestart,sothatprocessingofthesignalsisperformedonlywhentheycontainexcitation.Ifthestandarddeviationsofthesignalsrk,yk,uk,andvkintheactivebufferarebelowtheirthresholds,theexecutioniscancelled.2.2.3.CopyactiveMFMfromMIATheonlinelearningprocedurealwayscomparesthenewlyidentifiedlocalmodelswiththeprevioussetofparameters.Therefore,theactiveMFMiscopiedfromtheMIAwhereitisstored.AdefaultsetofmodelparametersisusedforthelocalmodelsthathavenotyetbeenidentifiedseeSection2.3.2.2.4.SelectlocalmodelsAlocalmodelisselectedifthesumofitsmembershipfunctionsbjkovertheactivebuffernormalisedbytheactivebufferlengthexceedsagiventhreshold.Onlytheselectedlocalmodelsareincludedinfurtherprocessing.2.2.5.IdentificationThelocalmodelparametersareidentifiedusingthefuzzyinstrumentalvariablesFIVidentificationmethoddevelopedbyBlazˇicˇetal.2003.ItisanextensionofthelinearinstrumentalvariablesidentificationprocedureLjung,1987forthespecifiedMFM,basedonthelocallearningapproachMurraySmithJohansen,1997.Thelocallearningapproachisbasedontheassumptionthattheparametersofalllocalmodelswillnotbeestimatedinasingleregressionoperation.Comparedtotheglobalapproachitislesspronetotheproblemsofillconditioningandlocalminima.Thismethodiswellsuitedtotheneedsofindustrialoperationintuitiveness,gradualbuildingofthenonlinearmodel,modestcomputationaldemands.Itenablesinventoryofthelocalmodelsthatarenotestimatedproperlyduetoinsufficientexcitation.Itisefficientandreliableinearlyconfigurationstages,whenalllocalmodelshavenotbeenestimatedyet.Ontheotherhand,theconvergenceinthevicinityoftheoptimumisslow.Therefore,itislikelytoyieldaworsemodelfitthanmethodsemployingnonlinearoptimisation.Thefollowingbrieflydescribestheprocedure.Modelidentificationisperformedforeachselectedlocalmodeldenotedbytheindexjseparately.TheinitialestimatedparametervectorhjMIAiscopiedfromtheactiveMFM,andthecovariancematrixPj,MIAisinitialisedto105Iidentitymatrix.TheFLSfuzzyleastsquaresestimates,hjFLSandPj,FLS,areobtainedusingweightedleastsquaresidentification,withbjkusedforweighting.Thecalculationisperformedrecursivelytoavoidmatrixinversion.TheFIVfuzzyinstrumentalvariablesestimates,hjFIVandPj,FIV,arecalculatedusingweightedinstrumentalvariablesidentification.Inordertopreventresultdegradationbynoise,agPractice–deadzoneisusedineachstepofFIVandFLSrecursive
编号:201311171115095776    大小:852.24KB    格式:PDF    上传时间:2013-11-17
  【编辑】
5
关 键 词:
教育专区 外文翻译 精品文档 外文翻译
温馨提示:
1: 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
2: 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
3.本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
5. 人人文库网仅提供交流平台,并不能对任何下载内容负责。
6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
  人人文库网所有资源均是用户自行上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作他用。
0条评论

还可以输入200字符

暂无评论,赶快抢占沙发吧。

当前资源信息

4.0
 
(2人评价)
浏览:29次
英文资料库上传于2013-11-17

官方联系方式

客服手机:17625900360   
2:不支持迅雷下载,请使用浏览器下载   
3:不支持QQ浏览器下载,请用其他浏览器   
4:下载后的文档和图纸-无水印   
5:文档经过压缩,下载后原文更清晰   

相关资源

相关资源

相关搜索

教育专区   外文翻译   精品文档   外文翻译  
关于我们 - 网站声明 - 网站地图 - 友情链接 - 网站客服客服 - 联系我们
copyright@ 2015-2017 人人文库网网站版权所有
苏ICP备12009002号-5