Fault Detection Method for Aero-engine Spindle Bearing Using an Adaptive Multiscale Morphological Filter

Aiming at the optimization of structural element scale of common multi-scale morphology (CMM) for identifying weak faults in rolling bears, an adaptive multi-scale morphology method is proposed. Firstly, difference operator based on morphological basic operators is established because it can extract bidirectional impulses. In order to define optimal ones in certain scale range that are sensitive to bearing fault feature, kurtosis is considered as basis for the selection, and the final fault signals are reconstructed by averaging the results of the optimized ones, which enables the comprehensive analysis of fault signals and the demodulation of fault characteristics. With vibration signals collected from rolling bearing with the outer and inner ring faults, it is validated that the proposed method is effective to diagnose the two bearing faults under strong noise background. Compared with CMM algorithm, the amplitudes of the extracted feature frequencies are nearly doubled, and it is suitable to identify the inner and outer ring faults of rolling bearings with weak signal-to-noise ratio.