Feature Extraction of Bearing Status Based on Multi-Scale Bistable Stochastic Resonance Array

摘要

Multi-scale bistable array (MSBA), which combines normalized scale transform, stochastic resonance effect driven by colored noise and parallel array, can be applied to weak signal detection under heavy noise. The experimental application in incipient fault feature detection of rolling element bearing has verified the effectiveness of MSBA model. This paper studies feature extraction method of rolling element bearing status degradation based on enhanced detection effect of MSBA model. Integrated features are proposed using local spectrum kurtosis and local signal-to-noise ratio of fundamental component of bearing faults. Different sizes of damages are planted on outer race of bearings for experimental validation.