自組織數(shù)據(jù)挖掘在高爐爐溫預測控制中的應用.pdf

1、浙江大學理學院碩士學位論文自組織數(shù)據(jù)挖掘在高爐爐溫預測控制中的應用姓名：賀詩波申請學位級別：碩士專業(yè)：運籌學與控制論指導教師：劉祥官20080501ABSTRACTSteelisoneofthemostimportantandwidelyusedmaterialsinmodernsocietyitisthebasisofsustainabledevelopmentofnationaleconomyAstheupperprocedurei

2、ntheironandsteelindustryBlastFurnace(BF)ironmakinghasgreatinfluenceonthesucceedingprocessofsteelmakingItisalsoaprimefactorfortheellcrgysavingandconsumptionreductionHoweversinceBFiron—makingprocessishighlycomplicatedwhose

3、operatingmechanismischaracteristicofnonlinearitytimelagllighdimensionbignoiseanddistributionparameteretc，itsmechanismisstillUBknOWllThecomplexityofthewholeprocessmakesitdifficultforautomaticcontr01Themostdifficultpartofa

4、utomaticcontrolforBFironmakingistomakeaccuratepredictionofsiliconcontentinhotmetalunderinstablestatusandBsethepredictedinformationtocontroltheironmakingprocess，whichisalsothefrontierofresearchinthefieldofmetallurgicalaut

5、omationThecunentworkusesdatacollectedfromBlastFurnaceNo6inBaotouSteel(with趾innervolumeof2500m3)assamplespaceAprofoundanalysisoftheexistingpredictivemethodsforsiliconcontentisexercisedToovercomethedeficienciesofexistingme

6、thods，aselforganizingdataminingmethodisproposedThismethodcombinesthemeritsofneuralnetworks，geneticalgorithmandregressionanalysisandCanchoosethemodelautomaticallyaftergivingtheinputsofsystemandmodelselectioncriterionAndit

7、sbiggestmeritisthatittakesinteractionbetweentheinfluencingfactorsintoconsiderationThroughiIld印thinvestigationtheinputvariablesthatcarrymostinformationofthesystemaswellastransferfunctionofmiddlenetworkandmodelselectioncri

8、terionareintroducedTheGMDHalgorithmisthenusedtoconstructamodelforpredictionofsiliconcontentinhotmetalandgoodresultsareobtainedTheinteractionbetweeninputvariablesishelpfulforanalysisofthesystemTofurtherexplorethealgorithm