基于多元統(tǒng)計過程控制的故障診斷技術研究.pdf

1、UDCY908728密級——諏牡工繁失港碩士學位論文英文題目盥￡絲￡hQⅡE型!!＆igD!i!￡￡地ig!!!堂!d!＆研究生姓名(簽名)!鹽巫指導教師姓名(簽名)!!丕!鬈職稱熬援申請學位學科名稱■擎摯掣彗夔＆呈土t赴學科代碼』8Q地論文答辯日期!QQ!：墊j學位授予日期2QQ：：嬰。。。哆j；六—芝≥學院負責人(簽名)二7評閱人姓名昱越評閱人姓名一～鏖室堊2006年6月1日湖北工業(yè)大學碩士學位論文AbstractMachinefa

2、ultdiagnosisisusuallyusedtoprovidescientificmethodologytodetectanddiagnoseproductionprocessfaults，findOutfaultsources，faultfrequencies，faultseveritiesandfaulttendenciesetc，andtakeeffeclivesolutionsFaultdiagnosisbasedonmu

3、ltivariatestatisticalanalysisiSanimportantbranchinfaultdiagnosisdomainsThecrucialarithmeticiSsystematicallyreseatchedthecurrentmethodsappliedinthepracticalprocessandtheirweaknessesarediscussedandtherespectivesolutionsare

4、introducedinthethesisThefollowingpartshavebeenstudied：(1)Tllecurrentrese￡／rchanddevelopingtendencyoffaultdiagnosisandstatisticalprocesscontrol(SPC、aresummarizedTheprimarymathematicaltoolsoffaultdiagnosisbasedonmultivaria

5、testatisticaltechnique，includingPrincipalComponentAnalysis(PCA)，PrincipalComponentRecursion(PcR)’andPartialLeastSquares(PLS)，andtheirapplicationsinfaultdiar鯛osisarebrieflyintroducedinthethesis(∞Thedevelopingmethodoffault

6、detectionanddiagnosissystembasedonPCAfortheproductionprocessiSsummarizedThemainstatisticalvariablesmultivariatecontrolchartsandtheanalysismeansofstatisticalvariablesdataareintroducedTheresultsoffaultdiagnosisbased0nPCAwi

7、thtwosimulationexamplesareanalyzed(3)ThealgorithmicprinopleofKernelPrincipalComponentAnalysis(KPCA)，whichcanbeusedtoreducethedimensionalityofnonlinearhigkdimensionaldataspaceandextractfeature，isresearchedItssimulationres

8、ultsindicatethatthemethodbasedon“KPCAcanactualizethebearingfaultclassificationaccuratelyItismoresuitablefornonlinearfeatureextractionfromthebearingfaultsignalsthanPCAH1Theessentialmethodsandthedisadvantagesinconditionmon

9、itoringandfaultdiagnosisofbatchprocesswithMultiwayPrincipalComponentAnalysis(MPCmareanalyzedAnon—linebatchprOcessmonitoringmethodwithaconsecutivelyupdatedMPCAmodeIiSproposedinthethesisTheproposedmethodcanbeusedtOeliminat

10、emanyfalsealarmsgeneratedbythefixedmodelandimprovethereliabilityofthemodelFinallythewholethesisissummarized，andsomefutureresearchareasofconditionmonitoringandfaultdiagnosistechniquebasedonmultivariatestatisticalprocessco