基于sk-nlm包络的滚动轴承故障冲击特征增强

《基于sk-nlm包络的滚动轴承故障冲击特征增强》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。

1、基于SK-NLM包络的滚动轴承故障冲击特征增强熊国良，胡俊锋，陈慧，张龙(华东交通大学机电与车辆工程学院，南昌330013)摘要：非局部均值算法(non-localMeans,NLM)是活跃于图像信号处理领域的一种新方法，因其良好的去噪特性，近几年来在滚动轴承故障诊断领域也开始获得应用。NLM利用样本点邻域窗口包含的局部结构为基本单元，通过对相似成分加权运算后取其平均值以达到抑制噪声T•扰、突出故障冲击特征的H的。但对于强噪声条件下的低信噪比信号而言，NLM滤波效果并不理想。提出一种结介谱悄度(spectralkKurtosis,SK)和NLM权重

2、包络谱的故障诊断方法，首先对原始信号进行SK分析得到最优中心频率及带宽构成最优滤波器，初步消除环境干扰及测量噪声;其次对NLM算法进行改进，不再以滤波信号为分析对象，而是直接利用NLM加权运算得到的信号样木点权值分布曲线作为预处理信号的包络信号，从权重角度使故障冲击得到二次削强，消除SK带通滤波器的带内噪声；最后对权值分布曲线进行包络谱分析，进而得到诊断结果。通过仿真信号、实验室信号及T•程应用信号分析对所提方法进彳亍了验证，并与最小爛解卷积(minimumentropydeconvolution,MED)进行了对比。关键词：滚动轴承；谱悄度；非局

3、部均值算法；加权运算；故障诊断中图分类号：TII165.3TN911.7文献标识码：A国家标准学科分类代码：460.4099FeatureEnhancementforRollingBearingsbySpectralKurtosisandNon-LocalMeansXIONGGuoliang,HUJunfeng,CHENHui,ZHANGLong(SchoolofMechatronics&VehicleEngineering.EastChina・JiaotongUniversity.Nanchang330013,China)Abstract:The

4、nonlocalmeans(NLM)isanewmethodwhichattractssignificantattentioninthefieldofimageprocessingasitovercomesthelimitationsoftheneighborhoodfiltereffectively.Todate,NLMhasbeenextendedtorollingbearingfaultdiagnosisduetoitsabilitytoreducethebackground.Unfortunately,thediagnosisresulto

5、fNLMisnotperfectinthecaseoflowSNR・Aimingatsuchproblem,ahybridapproachusingspectralkurtosis(SK)andNon-LocalMeanshasbeenproposedtodetectrollingelementbearingfaults・Theapproachconsistsofthefollowingthreemainsteps.Firstly,theoptimalcenterfrequencyandbandwidthisdeterminedbyspectral

6、kurtosisanalysisfortheoptimalfilterwhichhastheabilitytoeliminateenvironmentaldisturbanceandmeasurementnoise.AndthenthecyclicimpactfeatureisfurtherenhancedintheformofweightdistributioncurvebytheweightedoperationofNLM.Finally,thecharacteristicfaultfrequenciescouldbereadilydetect

7、edthroughthespectmmofweightseries.Theeffectivenessandmeritsoftheproposedapproachwereverifiedbynumericalsimulationandexperimentaldataofrollingelementbearingwithdifterentkindoffaults,withcomparisonwithminimumentropydeconvolution(MED).Keywords:RollingBearings;SpectralKurtosis;Non

8、localMeans;WeightedOperation;FaultDiagnosis［苛士节”z称。因工作环境通常十分恶