基于经验模态分解与神经网络的在线状态监测.pdf

《基于经验模态分解与神经网络的在线状态监测.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、机床与液压Dec．20l3HydromechatronicsEngineeringVoI．4lNo．24DOI：10．3969／j．issn．1001-388I．2013．24．010On-lineConditionMonitoringBasedonEmpiricalModeDecompositionandNeuralNetworkXIEFengyunSchoolofMechanicalandElectronicalEngineering，EastChinaJiaotongUniversity，Nanehang330013，ChinaAbstract：On-lineconditi

2、onmonitoringinmachiningprocessesplaysasignificantroletoimprovethemachiningstabilityandprecision．Inthispaper，anapproachbasedonempiricalmodede。composition(EMD)andneuralnetworkforon—lineconditionmonitoringisproposed．Therootmeansquare(RMS)ofintrinsicmodefunctions(IMFs)byEMDisregardedasmachiningpr

3、o‘cessingfeature．ThethreelayersBack—propagation(BP)neuralnetworkmodeltakingthema。chiningfeatureastargetinputofneuralnetwork，theIMFsascharacteristicparameter，andthe3typesofprocessingstatesasoutputareestablishedtoidentifytheprocessingstate．Theresultshowsthattheproposedmethodcaneffectivelyidenti

4、fythestateofofprocess．Keywords：empiricalmodedecomposition，neuralnetwork，conditionmonitoring，rootmeansquareOn—-lineconditionmonitoringinmachiningoper·-importantcontributionspresentedforconditionmonl—ationsisverycrucialinordertopreventtoolfailures，toring．f0rinstance．amethodofstaterecognitionsin

5、creasemachineutilizationanddecreaseproductionbasedonwaveletandhiddenMarkovmodelwaspres—costinanautomatedmanufacturingenvironment．On—entedbyXie[2]；anapproachformonitoringthellnesystemdiagnosticsandprognosticscanbeper—cuttingtoolconditionbyself—organizingfeaturemapsformedbyusingtherealtimemonit

6、oringdata[1]．(SOFM)waspresentedbyOwsley，etal[3]；AnewHowever，theon—lineconditionmonitoringisnothybridtechniqueforcuttingtoolwearmonitoring，aneasytaskforsomereasons，forinstance，thema—whichfusesaphysicalprocessmodelwithanartificialchiningprocessesareusuallynon—linear，andtime—neuralnetworks(ANN)m

7、odelisproposedforturn—variantsystems．whichmakethemdimculttobeingbySick[4]；Arealtimemonitoringmethodofmodeled；theacquiredsignalsfromsensorsarede—toolwearusingmultiplemodelingmethodwaspro—pendentonotherkindofmeasuringfactors，itisnotaposedbyErtu