Fault Feature Extraction of Rolling Bearing based on Improved EMD and Spectrum Kurtosis

Abstract

According to the modulation characteristic of fault signals of rolling bearings and the disadvantages of depending on the experience to select resonance high frequency band,an improved empirical mode decomposition（EMD） and spectrum kurtosis method of rolling bearing fault diagnosis is put forward.First of all,the bearing fault signal are decomposed into a number of intrinsic mode functions（IMF） through the EMD method.Then,the false IMF components is eliminated through mutual information,kurtosis and cross- correlation,the fault signal is reconstructed.Finally,the optimal band pass filter is designed by using the spectral kurtosis,then analysis of envelope demodulation spectrum of the filtered signal is carried out,the fault feature of rolling bearing is extracted.The analysis results of rolling bearing experimental signal show that,the improved EMD and spectral kurtosis method can effectively extract the fault features of rolling bearing,and has more advantages than the traditional envelope analysis method.

Keywords

• Empirical mode decomposition;Spectrum kurtosis;Rolling bearing;Fault diagnosis