火电厂厂级监控信息系统SIS建模、实现及人工智能的应用研究

《火电厂厂级监控信息系统SIS建模、实现及人工智能的应用研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、东南大学博士学位论文火电厂厂级监控信息系统（SIS）建模、实现及人工智能的应用研究姓名：余廷芳申请学位级别：博士专业：热能工程指导教师：林中达20040901垒!!!!!壁——AbstractInordertoimprovethewh01eeconomiebenefitsandthe1evelofproducemanagementforpowerplantinourcountry，theappearanceofSupervisoryInformationSystem(SIS)inplantlevelisinevitable，andtheproblemscorrelatedwithsis

2、systemhavealreadybecomeakindofhotsubjectfurresearch．Becausecomplexityofstudyobjcotsinpowerplant,nonlinear,couplingrelationamongparameters，duringtheprocessofestablishinganddevelopingofSISfunctionalmodelsexiststheproblemssuchasinaccurately,sometimeimpossibletosetupamodel．Inthisthesis，Usingthedevel

3、opmentof“The012-linesystemofunitperformancemonitoringandenergylossanalysisinpowerplant”andestablishmentofoptimizationproblemmodelsinSiSasexamples，conventionalmethodsofmodelingfursolvingtheproblemssuchasthecalculationofunitperformance，determinationoftargetvalueofoperationparameters，energylossanal

4、ysisastheoperationparametersdeviatesfromtargetvaluewereintroduced，andadvantagesanddisadvantagesofthosemethodswerealsodiscussed．Inordertoaddresstheaboveproblemsoftheconventionalmodelingmethods，theArtificialIntelligence(A1)modelingmethodssuchasArtificialNeuralNetwork(ANN)andGeneticAlgorithms(GA)we

5、readoptedintheSISfunctionalmodelestablishment，whichhasadvantageinthemodelingofcomplexobjacts．TheANNofSelfOrganizationFeatureMaprSOM)andBPnetworkwereadoptedtoestablishthemodelofsteamturbinepoweroutput，withtheaidofSOM’sabilityinclustering，thelimitationofconventionalcollectionoftrainingsampleswerea

6、ddressed．Withthemodelaftertrained，thedifferencebetweenthepredictedpoweroutputcalculatedbyourmodelandtheactualvaluegeneratedbysteamturbinelocatedin1．5percentage．mostofthemlocatedin1percentage．Withtheestablishedmodelandthedataofpowerneededincyclecoolingwatersystem．thebestvacuumvaIueofcondensercanb

7、edetermined．andenergylossresultbydeviationofsomemainoperationparametercanalsobedetermined，comparedwithconventionalmodels，themodelestablishedbyushastheadvantageofpertinencetotheactualequipmentsincethesampledatawerecollectedfr