高速列车万向轴动不平衡检测的EEMD-Hankel-SVD方法.pdf

《高速列车万向轴动不平衡检测的EEMD-Hankel-SVD方法.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、第51卷第1O期机械工程学报V_01．51NO．102015年5月JOURNAL0FMECHANICALENGINEERINGMay2O15DoI：l0．3901，JME．2015．10,143．__IL—Ej同速列车万向轴动不平衡检测的EEMD．Hanke1．SVD方法水丁建明林建辉赵洁(西南交通大学牵引动力国家重点实验室成都610031)摘要：针对聚合经验模式分解(Ensembleempiricalmodeldecomposition，EEMD)的等效滤波特性依然存在模式分量问频带重叠较大的根本缺陷，提出一种高速列车万向轴动不平衡动态检测的新方法。该方法的核心是对万向节安装机座的振动信号

2、进行EEMD分解得到基本模式分量，应用基本模式分量信号来构造Hankel矩阵，对该矩阵进行正交化奇异值(Singularvaluedecomposition，SVD)分解，以奇异值关键叠层作为奇异值的选择准则对信号进行重构，应用重构信号的傅里叶谱来检测高速列车万向轴的动不平衡，消除EEMD分解模式频带重叠对故障特征的淹没和混淆效应，提高了谱的清晰度，凸显了故障特征。应用万向轴动不平衡试验数据对该方法进行试验验证，结果表明，该方法能够有效检测万向轴动不平衡引起的故障特征和万向轴的固有振动特征，与纯EEMD方法相比，该方法在谱的清晰度和故障表征力上得到了显著提高。关键词：高速列车；万向轴动不平衡

3、；聚合经验模式分解(Ensembleempiricalmodeldecomposition，EEMD)；Hankel矩阵：正交化奇异值(Singularvaluedecomposition，SVD)；动态检测中图分类号：U211：U270DetectionoftheDynamicImbalancewithCardanShaftinHigh—speedTrainApplyingEEMD-HankeI-SVDDINGJianmingLINJianhuiZHAOJie(StateKeyLaboratoryofTractionPower,SouthwestJiaotongUniversity,Che

4、ngdu610031)Abstract：Anewmethodofdetectingdynamicimbalancewithcardanshaftinthehigh-speedtrainisproposedapplyingthecombinationbetweenensembleempiricalmodeldecomposition(EEMD)，Hankelmarxandsingularvaluedecomposition(SVD)contrarytOthealiasingdefectbetweentheadjacentintrinsicmodelfunctionsexistingintheE

5、EMD．ThevibrationaccelerationsignalsofgimbalaredecomposedthroughEEMDtogetthediferentintrinsicmodelcomponents．TheHankelmarx，whichisconstructedthrongthesingledecompositionmodelcomponent，isorthogonallyexecutedthroughSVD．Thekeysingularvaluesareselectedtoreconstructvibrationsignsonthebaseofthekeystackofs

6、ingularvalues．Fourierspectrumofthereconstructedsignalisappliedtodetectdynamicimbalancewithshaftandeliminatesclutterspectrumcausedbythealiasingde~ctbetweentheadjacentintrinsicmodelfunctions，highlightsthefailurecharacteristics．Themethodisverifiedbytestdataintheconditionofdynamicimbalance．theresultssh

7、owthismethodcalefectivelydetectthefaultvibrationcharacteristicscausedbycardanshaftdynamicimbalanceandextractthenaturevibrationfeatures．WithcomparisontothesimpleEEMD，theclarityandfailurecharacterizationforce