Research on mathematical morphological operators for fault diagnosis of rolling element bearings

Morphological filtering adopting the combined morphological operators (CMOs) has been widely used to extract bearing fault features from vibration signals. However, few studies focus on the comprehensive performance of various CMOs under different interferences. In this paper, several new CMOs for feature extraction are proposed firstly, and then the impulse extraction performance of fourteen typical CMOs in the presence of harmonic interference, random impulses and background noise is investigated through simulations. To enhance the capability of impulse extraction and noise elimination, the morphological hat cross-correlation operator (MHCCO) is constructed through the cross-correlation of two CMOs with excellent performance. Additionally, an improved strategy is proposed to adaptively determine the optimal length of the structural element for MHCCO. Simulations, experiments and comparisons demonstrate the effectiveness and superiority of the proposed method. This paper provides important guidance for selecting CMO for feature extraction and an effective method for bearing fault diagnosis.