Fault Diagnosis of Rolling Bearing based on Optimal Resonance Sparse Decomposition

The method of resonance sparse decomposition (RSSD) is extensively used in rolling bearing fault diagnosis. The selection of the decomposition parameters plays a decisive role in fault separation. It is difficult to accurately diagnose the weak fault of rolling bearing by traditional methods. In this paper, the fault diagnosis method of the rolling bearing is performed based on signal resonance sparse decomposition. The resonance sparse decomposition is carried out according to the different quality factors (QF) of the harmonic component and the periodic impact component in the rolling bearing fault vibration signal. The decomposition effect of the signal resonance sparse decomposition method is closely related to the quality factor. However, the quality factor selection based on human experience is often not effective, and the interpretability is not strong. To ensure the accuracy of the parameter selection, this paper proposes a multi-parameter optimization method based on the Grey-Wolf optimization algorithm (GWO) for adaptive resonance sparse decomposition. The simulation test and application example show that this method can effectively extract the fault characteristic components of the bearing, eliminate the signal interference and noise, and correctly identify the fault state of the rolling bearing.