基于多振動信號信息融合的旋轉(zhuǎn)機械故障診斷方法.pdf

1、湖南科技大學(xué)碩士學(xué)位論文基于多振動信號信息融合的旋轉(zhuǎn)機械故障診斷方法姓名：李萍申請學(xué)位級別：碩士專業(yè)：機械電子工程指導(dǎo)教師：李學(xué)軍胡小平20080401湖南科技人學(xué)傾Ij學(xué)位論殳AbstractRotatingMachineriesareusedinindustriaJmachinerywidelyOneetheequipmentsarcfailureofrunningitwillcauseenormouseconomiclossesW

2、ithrotatingmachinerydevelopingtowardshigh—speed，high—power，highreliabilityandlarge，theinformationneedtobeconsideredismoreandmoreandthedif1]cultyofextractinginformationeffectivelyiSincreasedbecauseofthemutualinterferenceb

3、etweenSignalsandsignalssignalsandnoisesSOthetraditionalfaultdiagnosistechnologyhavefailedadaptingtocomplicatedequipmentfaultdiagnosisBasedonthisconsiderationthesignalintegrationtheorywasusedtofusetheusefulinformationtoen

4、hancetheaccuracyandthercliabilityoffaultdiagnosisInthisPaper，researchonintegrationandfaultdiagnosisidmadefromthethreelevelsofvibrationsignalsbasedoninformationtheoryandtechnologyintegrationOnthisbasisadetailedDescription

5、ofthemethodstoeverysignalintegrationlayers’achievingaremade1ThefaultdiagnosismethodofMulti—sensorsvibrationsignal’SdatafusionbasedorlcorrelationAsignal’SdatafusionbasedoncorrelationfunctioniSproposedfordata1evelintegrati

6、ontoimprovethesignalaccuracyfault—tolerantInconjunctionwiththeexamplCSwereanalyzedThroughtocarryonthepracticalapplicationinrevolvesdiagnosis，indicatedmatthismethodiSfeasibleandeffectiveinthepracticalapplication2Thefaultd

7、iagnosismethodofcharacteristicsinformationfusionbasedonneuralnetworktechnologyTheprocessofnetworkdesigniSdescribedindetailAmethodcombiningthewaveletpacketdecompositionandEmpiricalModeDecomposition(EMD)isusedtoextractthef

8、eatureofthesignalsFirstsignalwasdecomposedintoaseriesofnarrowbandsignalsusingwaveletpacketdecompositionandfeaturesignalsiSfallonthedifferentfrequencybandgenerallyThenvariousbandsignalsgottenfromwaveletpacketdecomposition

9、aremadefurtherdecomposedusingEMDdecompositionandsignalcharacteristicsareextractedfromtheIntrinsicModel：unctionfIMFlcomponentwhichthefeaturesignalsarecontained1“heextractionofthesignalfeaturesareputintotheneuralnetworksse

10、parately，andthenthefeature—signalfusionresultsaregainedThedetailedprocessdescriptionismadethroughacombinationofspecificexamples3111efaultdiagnosismethodofdecision—makinginformationfusionbasedonD—SevidencetheoryThedetaile

11、ddescriptiononthebasicconceptstheoryofD—SevidencetheoryandtheintegrationandreasoningmethodsofDSevidenceismadeinthispaperAdecisionmakinglevelsignalfusioniSsetuptomakeafurtherintegrationbaseontheDiagnosisofneuralnetworkand