A Hybrid Approach Based on the SR-HWPT-PDF for Identifying Early Fault Signals in Rolling Bearings

Abstract

An approach combining the stochastic resonance, the harmonic wavelet packet transforms and the probability density function was proposed to obtain the early fault signal of a rolling bearing. Firstly, an adaptive variable-scale stochastic resonance was employed to detect the frequency range of the rolling bearing’s fault signal based on the improved signal-to-noise ratio, and then the harmonic wavelet packet transforms and the probability density function were utilized to extract and identify the fault signal of the rolling bearings from the detected signals. The case studies show that the proposed method can effectively obtain the early fault signals of the outer race, inner race and rolling element in the rolling bearings despite not knowing the frequency band distribution, and that the early fault diagnosis of the rolling bearings can be achieved.

Keywords

• rolling bearings
• stochastic resonance
• harmonic wavelet packet transforms
• probability density function