Fault Feature Detection of Rolling Bearing Based on LMD and Third-Order Cumulant Diagonal Slice Spectrum

Abstract

Bearing fault signal presents obvious nonlinear and non-stationary characteristics, and local mean decomposition (LMD) method can adaptively process nonlinear and non-stationary signal. Quadratic phase coupling is one of the commonest nonlinear phenomena, and three-order cumulant diagonal slice spectrum is suitable for the extraction of the frequency components involved in quadratic phase coupling. In order to effectively detect the fault features of rolling bearing, a fault feature detection method of rolling bearing based on LMD and diagonal slice spectrum was proposed in the paper. Diagonal slice spectrum was applied in the extraction of the frequency components involved in quadratic phase coupling from the simulation signal, and it successfully extracted the quadratic phase coupling frequency components of the , and . The method was finally used in the detection of the bearing inner race fault features, and the results demonstrated that compared with diagonal slice spectrum of the PF component signals, the diagonal slice spectrum of the PF component envelope signals can identify fault characteristics much clearer and brighter. Meanwhile, the effective extraction of the inner race fault features verified the effectiveness and practicability of the method proposed in this paper.