Research on a bearing early fault features extraction method

In view of problems that early fault signals of rolling bearings are submerged by background noise and fault characteristics are not obvious, a bearing early fault feature extraction method based on wavelet packet decomposition and CEEMD was proposed. Matlab software is used to perform rapid spectral kurtosis analysis on the collected vibration signals, and the center frequency and bandwidth of the band-pass filter is determined according to maximum kurtosis principle and used to design the band-pass filter. Wavelet packet decomposition and CEEMD decomposition are perform to the signal filtered by the band-pass filter, and effective intrinsic modal function (IMF) components are selected according to the kurtosis and correlation coefficient and used to reconstruct the wavelet packet signal. Characteristic frequency of bearing early fault signal is extracted by envelope spectrum analysis of the reconstructed wavelet packet signal. The method reduces background noise interference through spectral kurtosis analysis, enhances the fault impact signal through wavelet packet decomposition, and combines CEEMD with wavelet packet decomposition to solve the problem of modal aliasing and invalid components in classical EMD decomposition. The simulation results show that compared with traditional envelope demodulation algorithm, the background noise of the reconstructed signal is suppressed and the fault feature component is prominent, which verifies the feasibility and effectiveness of the proposed method.