Research On Fault Diagnosis Method Of Motor Rolling Bearing Based On Improved Empirical Wavelet Transform

Motor is an important rotating machinery in modem industry.Rolling bearing,as the core component of motor,plays an extremely important role in the industrial field.Therefore,fault diagnosis of rolling bearing has important practical significance.Fault signals of rolling bearings are non-stationary and non-linear noisy signals.At present,time-frequency analysis is widely used in fault diagnosis of rolling bearings,such as windowed Fourier transform,Wigner distribution,wavelet transform,empirical mode decomposition(EMD)and so on.Empirical Wavelet Transform(EWT)is a new adaptive signal decomposition method.This method inherits the advantages of EMD and wavelet analysis and overcomes the endpoint effect and modal aliasing in EMD to a great extent.Therefore,applying the empirical wavelet transform method to the fault diagnosis of rolling bearings has important significance and theoretical value.This paper introduces the empirical wavelet transform theory.By comparing the experimental results with the empirical mode decomposition(EMD)method,the results verify the advantages of the modal decomposition of the empirical wavelet transform.By combining the other methods to decompose the bearing fault signal,an improved EWT based motor bearing fault feature extraction method is proposed.The improved empirical wavelet transform(FCMEWT)applies the fuzzy C-means and correlation coefficient method to the bearing fault feature extraction method.The fuzzy C-means clusters a set of maximum and minimum length curves in the scale space plane to obtain a set of spectrum division points.The correlation coefficient method is used to obtain a suitable threshold to select the decomposition modes,and finally a set of characterization bearings can be obtained.The IMF component of the fault feature achieves the purpose of bearing fault feature extraction.In order to solve the cumbersome steps in the above proposed method,the operation time is complicated,which is not conducive to practical engineering applications and the like.Through the study of scale space theory,it is found that the scale space transformation can make the fault characteristic signal appear,so this paper proposes an improved EWT method(MSCEWT)based on the maximum-minimum length curve.The method aims to improve the scale space plane.Through the convolution operation of the spectrum and the Gaussian kernel function at different scales,the spectrum division point is finally obtained.The power spectral density of the IMF component is obtained,and the fault characteristic frequency of the signal is obtained.The MSCEWT method is compared with the results of EMD and collective empirical mode decomposition(EEMD).The experimental results show that the improved method can effectively extract the fault characteristic frequency of the inner ring,the outer ring and the rolling element of the bearing and the number of decomposition modes is small,which can overcome the problems of modal aliasing and end effect.Therefore,this study provides a new method for bearing fault diagnosis.