Fault diagnosis of rolling bearing based on CEEMDAN reconstruction and fast spectral kurtosis

Abstract:
Aiming at the problem that the vibration signal of rolling bearing in wind turbine contains strong background noise, so it is difficult to analyze the fault characteristics, a bearing fault diagnosis model based on CEEMDAN decomposition and reconstruction and auxiliary fast spectral kurtosis graph algorithm is proposed. first, The signal is decomposed into multimodal components by CEEMDAN (IMF) and its margin, the effective IMF component is selected by using the kurtosis value and correlation coefficient criterion for signal reconstruction. Then, the fast spectral kurtosis graph algorithm is used to filter the reconstructed signal. Finally, the envelope spectrum is compared and analyzed to obtain the accurate fault characteristic information of rolling bearing. The research results show that the filtered signal is obtained by CEEMDAN decomposition and fast spectral kurtosis The reconstructed signal has high signal-to-noise ratio and can obtain more accurate fault bandwidth and center frequency. This method can effectively measure the vibration frequency of inner ring failure of 6205-2RS JEM SKF deep groove ball bearing.