Fault detection for rolling element bearing based on repeated single-scale morphology and simplified sensitive factor algorithm

A hybrid of repeated single-scale morphological filtering (RSSMF) and simplified sensitive factor (SSF) method is proposed to detect the fault signals of rolling element bearing. First, unit scale (three sampling points) in the morphology filtering is introduced to retain more feature components of a signal. To obtain a satisfied effect in morphological filtering, a repeated morphological differential operator (RMDO) is developed to perform in the RSSMF. After the repeated morphological filtering is implemented, a series of outputs are achieved. Some of them comprise interested information and others contain irrelevant one. To highlight useful information, some factors that are sensitive to the useful information are computed by the simplified sensitive factor algorithm. Finally, the reconstructed signals are obtained by the weighting sensitive factors. The proposed method is assessed by both simulation analysis and vibration signals of the rolling element bearings with the outer and inner race faults. Compared with traditional single-scale morphological filtering (TSSMT) and traditional multi-scale morphological filtering (TMSMT), the results demonstrate that the proposed approach has superior performance in noise removal and fault feature detection.