Rolling Bearing Fault Research based on Multiple Denoising and PSO-MCKD

Abstract

A bearing fault diagnosis method based on CEEMDAN-WPD-PSO-MCKD is proposed for the problem that the rolling bearing signal is susceptible to strong noise interference and difficult fault diagnosis in the operating environment. First, the method decomposes the rolling bearing signal into several Intrinsic Mode Functions (IMF) using the complete ensemble empirical modal decomposition of adaptive noise (CEEMDAN). Secondly, the kurtosis index of each order IMF is calculated, and the key IMF components are selected for signal reconstruction. The improved wavelet packet threshold function proposed in this paper is used to reduce the noise, and the sample entropy is used as the index to select the best threshold. Finally, the denoising signal is deconvoluted with the Maximum correlation kurtosis deconvolution (MCKD) optimized by the PSO algorithm to extract the fault frequency components in the best deconvoluted signal for fault diagnosis. The method has been verified by the XJTU-SY rolling bearing experimental data set, the results show that the method can effectively extract the fault features in the signal and has good fault identification effect.