工業(yè)過程噪聲的混沌特征研究.pdf

1、浙江大學博士后學位論文工業(yè)過程噪聲的混沌特征研究姓名：郜傳厚申請學位級別：博士后專業(yè)：系統(tǒng)工程指導教師：錢積新20060510浙江大學博士后研究工作報告ABSTRACTThemainreasonthatmanyclassicalcontroltheoriesareincapableofbeingappliedtopracticesucessfullyisthewrongmeanstodealwithnoiseCandidateexamp

2、lesinculdeKalmanfilterleastsquareidentificationmultivariatestatisticalanalysisneuralnetetcTheseabovementionedtheoriesandmethodsworkbasedontheassumptionofwhitenoiseorstationaryrandomprocessatmosLHoweversincewhtienoise西jus

3、tanidealmodel脅mathematicsandexistshardlyinindustrialprocesssuchassumptionistooutopiantobeappliedtopracticesuccessfullyThereforeitisnecessarytoexploreFlewmechanismtoexplainthenoisefromindustrialprocessNFIP)butnotfollowing

4、theassumptionofwhitenoisealwaysBecausetheresearchonnonlineardynamicsindicatesthatmanyphenomenawhichappeartoberandomarenotrandominherentlybutchaot缸nmotivatesUS雷investigateNFIPfromthepointofviewofchaosInthisreportthneserie

5、sdataoftemperatureactuallyobservedfromdehydrationtowerwithintheoxidationprocedureofproducingPTA(Purifiedterephthalicacis)inYangziPetrochemicalLtdistokenastheoriginalsampleThenweadoptasimpleapproachthatthemeanvalueofsucce

6、ssivemultipoints括三subtractedfromoriginaldatatoextraCtnoisebasedonthefollowingtwoassumptions：(1)theNFIP缸zeromean；(2)theeffectivesignal缸slowvaryingwhiletheNFIPisfastvi，ryingAfterwardscorrelationdimensionasanindicatorofchao

7、sisestimatedusingthetechniqueofphasespacereconstructionandGPalgorithmTheresultsshowthattheNFIPishighdimensionalchaoticsystembutnotrandomsystemandsayingnothingofwhitenoiseInadditionthenumberofdegreesoffreedomneededtomodel

8、thedynamicsoftheNFIPisobtainedThisreportpresentsawaytowardsdirectsolvingthepredictionandmodelingoftheNFlPwhichwillbreaktherigorouslimitationofwhilInoiseneededbyclassicalparameteridentificationKalmanfiltermultivariatestat