采用工作振动响应的机械结构的健康监测方法的研究

太原理工大学硕士学位论文采用工作振动响应的机械结构的健康监测方法的研究姓名：赵常举申请学位级别：硕士专业：机械电子工程指导教师：熊诗波20100501IToeplitz(SVD)II()HankelQR()LMSTest.labmatlabPolymax()PolymaxIIIPolymaxIVRESEARCHONHEALTHMONITORININGMETHODSBASEDONTHEWORKINGVIBRATIONRESPONSEOFTHEMECHANICALSTRUCTUREABSTRACTAtpresent,thestructuralhealthmonitoringmethodsbasedonstructuralvibrationresponsehasbeenfocusedbythescientificresearchinstitutionsathomeandabroad,andtheoperationalmodalanalysismethodsaremostvaluableamongthem.Duetotheoperationalmodalanalysismethodsusingonlythestructuralvibrationresponsedatawhichisobtainedinoperatingconditionsforon-lineanalysis,anditdoesnotrequireartificialmotiveequipmenttoincentivethestructure,sotheanalyzedresultsofsuchmethodsmoreinlinewiththeoperatingboundaryconditionsofthestructure.Therefore,suchanalysismethodshavebeenwidelyusedintheaerospacebridgesandmechanicalmanufacturingengineeringapplications.Forthesereasons,thestochasticsubspaceidentificationmethodrelyingonenvironmentalexcitationhasbeenexploringlystudiedonthispaper.DuringtheVresearchprocess,wehavemainlydonethefollowingworks:Onthepaper,wehavestudiedcantileverbythefiniteelementanalysismethodtoconducttheirtheoreticalmodalanalysis,andobtainedtheirmodalfrequenciesandmodalshape.Thusitprovidesatheoreticalframeofreferencefortheexperimentalanalysis,andcomparativelyanalysistheexperimentalmodalanalysisresultsbythestochasticsubspaceidentificationmethodandthetheoreticalresults,anditshowsthatthetwomodalidentificationresultsareingoodagreement.onthebasisofpreviousstudies,thestochasticsubspaceidentificationmethodwhichisinspiredbytheenvironmenthasbeenstudieddeeplyonthispaper.Thestochasticsubspaceidentificationmethodmainlyconsistsoftwokindsofalgorithms—thecovariance-drivenstochasticsubspaceidentificationmethodanddata-drivenstochasticsubspaceidentificationmethod,andthespecificidentificationprocessofthetwoalgorithmshasbeenrespectivelydiscussedindetailonthispaper.TheformermainlyusestheoutputcovariancesequencetocomposeToeplitzmatrixblocks,bythesingularvaluedecomposition(SVD)toobtainthesystemmatrixandcontrolmatrix,thenitusesthemtoidentifythemodalparametersofvibratingsystem(modalfrequenciesmodaldampingratioandthemodalshape);whilethelattermainlyusesHankelmatrixcomposedofthetime-domainvibrationresponsedata,reducestheamountofthematrixcalculationsbyQRdecomposition,andeliminatestheeffectofnoisebythesingularvaluedecompositionoftheprojectionmatrix,thenitusesthematrixVIoperationssuchasleast-squaresmethodtoidentifythesystemmatrixandoutputmatrix.Finally,itusesthesystemmatrixandoutputmatrixtoidentifythemodalparametersofvibratingsystem(modalfrequenciesmodaldampingratioandmodalshape).Becauseofresearchinglessonthesystem-orderdeterminationmethodbasedonthestochasticsubspaceidentificationmethod,weconductedanexploratorystudy.Onthispaper,wemainlystudiedthesystemorderdeterminationmethodbythesteady-statediagramorsingularentropyincrement,whicharebothbasedonthestochasticsubspaceidentificationmethod.Theformerhasbeenon-linecompletedbyLMSTest.labmodalanalysissoftware,butthelatterhasbeencompletedbythematlabapplication.Theresultsshowthattheresultsoftwoorderdeterminationmethodsareingoodagreement,andtheyaremorereliableandeffectivethanthetraditionalsystemfixed-ordermethodwhichusesthesingularvaluedecomposition.Aftercomparingwiththetheoreticalresultsobtainedbythefiniteelementanalysismethod,westilltakethecantileverstructureasresearchedobjectsandincentiveitbyarandomwhitenoisesignaltomimicitsworkingstateintheenvironmentalconditions,withtherandomsubspaceidentificationmethodandPolymax(Least-SquaresComplexFrequency-domainmethod)conductingon-linemodalanalysisforthem.Theresultsfurtherprovesthevalidityandreliabilityofthestochasticsubspaceidentificationmethod.Insummary,thefiniteelementmodalanalysisandPolymaxmodalanalysisVIImethodareadopted,andtheyrespectivelyprovedthevalidityandreliabilityofthestochasticsubspaceidentificationmethod.Inaddition,onthesystemorderdeterminationmethods,westudiedtwokindsofsystemorderdeterminationmethodsbasedonthestochasticsubspaceidentificationmethod—thesystemsteady-statediagramfixed-ordermethodandsystemsingularentropyincrementorderdeterminationmethod,theresultsshowedthatthefixed-orderresultsoftwosystemorderdeterminationmethodareingoodagreement.Inshort,becausethestochasticsubspaceidentificationmethodisakindofmodalidentificationmethodbasedonthestructurevibrationresponse,andtheidentifiedresultsisstableandreliable,andcurrentlythesystemfixed-ordermethodbasedonthestochasticsubspaceidentificationmethodhasbeenlessstudied,soitplaysanimportantroleinthemechanicalOn-linehealthmonitoringstudies.KEYWORDS:structuralhealthmonitoring,stochasticsubspaceidentificationmethod,Polymax,steady-statediagramorderdeterminationmethod,singularentr-opyincrementorderdeterminationmethod11.1[1][2]Polymax[3][4]QRSVD[5]AC21.220096121612A33032009715-1541531520101097.5RAV4200346[6]1.331988PENVAL2002-20032080NASA(uTRe)Roeketdyne2004LMSTest.labAleniaSpazioLMSSYSNOISE125Hz19ColumbusEADSInternet()3S(GISGPSRs)1.442070ITDPolymax[7][8]Cholesky[9][10][11]PaulWilliams[12][13]