基于CEEMDAN和FastICA的轴承故障诊断.pdfVIP

机械工程师

MECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEER

基于CEEMDAN和FastICA的轴承故障诊断

卞东学,李甲历

（沈阳化工大学机械与动力工程学院，沈阳110142）

摘要：

针对滚动轴承早期故障特征信号微弱、难以提取的特性，引入一种基于总体经验模态分解方法（CEEMDAN）和快

速独立分量分析（FastICA）相结合的滚动轴承故障特征提取方法。该方法首先通过CEEMDAN和峭度、相关系数结合，完成

信号的分解与重构。然后构建噪声通道，使用快速独立成分分析进行去噪分析，获得去噪信号。最后，对分离出的最佳估计

信号进行包络谱分析并得到故障特征频率。该方法有效降低了噪声干扰，能够对故障特征频率进行有效识别。

关键词：

特征提取；总体经验模态分解；快速独立分量分析

中图分类号:TH133;TG506文献标志码:粤文章编号：员园园圆原圆猿猿猿（圆园22）0原园044原园4

BearingFaultDiagnosisBasedonCEEMDANandFastICA

BIANDongxue,LIJiali

(SchoolofMechanicalandPowerEngineering,ShenyangUniversityofChemicalTechnology,Shenyang110142,China)

Abstract:Aimingatthefeaturethatearlyfaultfeaturesignalsofrollingbearingsareweakanddifficulttoextract,a

methodforextractingfeaturesofrollingbearingfaultsbasedothecombinatioofcompleteensembleempiricalmode

decompositiowithadaptivenoise(CEEMDAN)andfastindependentcomponentanalysis(FastICA)isintroduced.This

methodcombinesCEEMDANwithkurtosisandcorrelatiocoefficienttocompletethedecompositioandreconstructioof

thesignal.Thethenoisechaelisconstructed,andthefastindependentcomponentanalysisisusedfordenoising

analysistoobtaithedenoisingsignal.Finally,theseparatedbestestimatiosignalisanalyzedbyenvelopespectrum,and

thefaultcharacteristicfrequencyisobtained.Thismethodeffective