A performance enhanced time-varying morphological filtering method for bearing fault diagnosis

Fault feature extraction and broadband noise elimination are the keys to weak bearing fault diagnosis. Morphological filtering is a typical fault feature extraction method. However, the parameter selection of structural element (SE) has an important influence on the filtering result. To solve this problem, an adaptive time-varying morphological filtering (ATVMF) is proposed. ATVMF adaptively determines the shape and scale of SE according to the inherent characteristics of vibration signal, effectively improving the fault feature extraction capability and computational efficiency. To solve broadband noise pollution, the diagonal slice spectrum (DSS) is applied to the resulting signal of ATVMF to further eliminate the fault-unrelated components. Finally, a weak bearing fault diagnosis method combining ATVMF and DSS is developed. Simulation and experimental results verify that the proposed method can effectively enhance fault-related impulse features and diagnose weak bearing faults. The comparison with several existing methods demonstrates the advantages of the proposed method.