Signal complexity analysis for fault diagnosis of rolling element bearings based on matching pursuit

Abstract

Various techniques have been presented to detect rolling element bearing faults, however, it is still a huge challenge to accurately extract features under high noise level. In this paper, a new approach based on matching pursuit for analyzing the signals of rolling element bearings is proposed. Different from most of the matching pursuit related works focusing on the time-frequency plane, complexity spectrum and complexity spectrum entropy are proposed in this research to accurately extract the fault feature of a rolling element bearing signal under a low signal to noise ratio. Both simulation and experiment show that complexity spectrum works better than envelope spectrum in distinguishing characteristic frequencies of fault bearings submerged in noise. An accelerated whole lifetime test of bearing has been performed to collect vibration data, which is analyzed by complexity spectrum entropy and other normal approaches. Results show that complexity spectrum entropy has some unique characteristics.