The diagnosis approach for rolling bearing fault based on Kurtosis criterion EMD and Hilbert envelope spectrum

Abstract

A fault diagnosis approach based on empirical mode decomposition(EMD), Hilbert envelope spectrum and Kurtosis criterion is proposed, by considering the nonlinear and non-stationary characteristic of fault signal for rolling bearing. The fault signals are decomposed into a sum of intrinsic mode functions(IMFs) with EMD decomposing process. Also, some of the IMFs are selected to reconstruct the signal through using Kurtosis criterion, which can be used to represent the fault information. And then, the determination of specific fault for rolling bearings are achieved by comparing the envelope spectrum of the reconstructed signal with the fault characteristic frequency. It can be found from the result that the proposed approach can be well used in the fault information extraction for the rolling bearing, and with a good application prospect.