Empirical mode decomposition and Hilbert spectrum analysis based bearing faults diagnosis

Abstract

This paper presents a new method to analyze the vibration signal through using empirical mode decomposition and Hilbert spectrum analysis. This method solves a problem of the time domain analysis to detect the bearing fault types and traditional resonance demodulation to select the band-pass of the band-pass filter. First of all, a vibration signal is decomposed into a number of intrinsic mode functions, and then the original vibration signal correlation in the intrinsic mode functions for Hilbert spectrum analysis to get its envelope spectrum and extract the fault characteristic frequency from the vibration signal. Finally, using the data of Case Western Reserve University in the United States, the method is used to verify the bearing status monitor and its fault types. Experimental results show that the new method can diagnose the bearing three kinds of surface defects early in the outer ring and the inner ring fault correctly.