基于改進ESN的高爐煤氣系統(tǒng)預(yù)測方法的研究.pdf

1、大連理工大學(xué)碩士學(xué)位論文基于改進ESN的高爐煤氣系統(tǒng)預(yù)測方法的研究姓名：時飛飛申請學(xué)位級別：碩士專業(yè)：控制理論與控制工程指導(dǎo)教師：王偉20091201基丁改進ESN的高爐煤氣系統(tǒng)預(yù)測方法研究StudyonthePredictionMethodforBlastFumaceGasSystemBasedonIⅡlproVedEchoStateNe觚orkAbstractThe仃msportation鋤dConVersionprocessofe

2、11ergyinsteelindustryisrelativelycomplicatedItis鋤importanttaskfore11te印riseen錨gymanagementtoguaraIlteeacontinuous，safe鋤deConomicalene曬，supplyAtpresent，mebyproductgasismekeypaItofeIlergyoptimalschedulingprobl咖Blast如maCeg嬲

3、(BFG)iswimthecharactI斑sticsofle嬲tcaloricityComplexgenerationpmcessaIldoutputwimbigfluctuation，soitwillusuallybedi朐sedfirstlywhenpoorenergyschedulingandimbalaIlceoccurs，、他ichwillcreatemuchmorehann如lg嬲andde黟adet11eenVironm

4、髓tThereforc，thestudyonreliableprediCtionofnleBFGgeIlerationa11dconsumptionanditsrecover)ra11drcuse如llycanprovidescielltificguidaJlceforemcientutilizationofthebyp∞ducteIlergyaIldimproVemele、，elt0saVeen鶘yinsteelente印riseII

5、l“spaper，onbasisofmebackgroundofShaIl曲aiBaosteelthegeIl訓(xùn)ionaIldconsumptionusersoftheBFGsystenlarediVidcdintotwotypesformepredictionproblem，whichareadjustableusersaIldnonadjustableu8ers，respectivelyFornonadjustableusers，趾

6、improVedechostatenetwork(ESN)predictionmetllodwimpar鋤etersispmposedfirstlyBaSedonLe嬲tMe鋤Square(LMS)e玎0f碰t嘶on，theIlel[、)l，orkpar鋤etersareoptimizedbystochastic伊adieIltdescentinordertoachieVea111iIlimized仃aillingenorComp卸ed

7、toothermethods，suchmethodbdngsaboutabetterprcdictionperfomallceSeCondly，a11averagemethodisusedtopredictadiustableusersSincemedataoftheBFGgenerationincludesas耐esofnoisesinpractice，tllep耐omlallceisu11satisfaIcto巧ifusingthe

8、iIIlprovedESNmethoddirectlyForresolVingthisproblem，aIlapproachbaSedon鋤piricalmodedecomposition(EMD)isemployedtoimpleIllentthedatadenoisebeforepredictionBasedona11alyzingtheinfluencefIactorSofgasholderlevel，predictionfact

9、ormodelofmeBFGgasholderisbuiltbytlleESNmethodAn印plicationsoR、張re南rmeBFGsysterIlpredictionisdevelopedonthebasisofthepredictionmemodTtlisso仕wareconsistsofserveraIldclienttennillaLlTheserverpredictstheBFGsystemwllichinclude

10、sallusers肌dgash01deratregularinterValsalldtheclienttenninalpresents也erealtimedataa11dp捌ictdatathrou曲theuscrinte矗犯esThepIedictionsof研arehasbeen印pliedintlleEnergyCenterofBaosteela11dthenmnjngresultsd鋤onstratethevalidityoft