基于ICA包络增强MEMD的滚动轴承故障诊断

针对多元经验模态分解(MEMD)存在模态混叠、带内噪声干扰导致轴承故障特征信息微弱难提取问题,提出了基于独立分量分析(ICA)包络增强MEMD的滚动轴承故障诊断。采用MEMD对多通道信号进行自适应分解,依据峭度和相关系数选取包含故障信息的本征模态函数(IMF);对所选取IMF分量的包络信号进行ICA分析,抑制模态混叠和削弱带内噪声;选取峭度最大的独立分量包络进行频谱分析,判断滚动轴承的运行状况。实测信号结果表明:ICA包络增强MEMD后包络谱中可以清楚地看到前6阶故障频率,故障特征频率误差小于1 Hz,其他方法只能看到2~3阶,且干扰频率成分较多。 

Rolling bearing fault diagnosis based on MEMD with ICA envelop enhancement

In view of existing problem of mode mixing and in-band noise of the intrinsic mode function(IMF) after multivariate empirical mode decomposition(MEMD), a rolling bearing fault diagnosis method based on MEMD with independent component analysis (ICA) enhancement was proposed. multi-points vibration data were adaptively decomposed, and a series of IMF were obtained by MEMD, then optimal IMF component was chosen by kurtosis and correlation coefficient. The envelop waveform of the chosen IMF were fed to ICA to restrain mode mixing and weaken in-band noise. The best ICA component was selected by max. kurtosis of envelop, and its spectrum was used for bearing diagnosis. The experimental results showed that the first 6 order characteristic frequency of enhanced MEMD spectrum by ICA can be seen, and the frequency error was less than 1 Hz. The other methods with more interference frequency components only allow to see 2-3 orders.