Improved RSSD and Its Applications to Composite Fault Diagnosis of Rolling Bearings

Due to the influences of transmission paths and various interference sources, the individual defect-induced fault features of bearings simultaneously arising from multiple defects were difficult to extract from vibration signals, an improved RSSD method was proposed, which was combined with dual-parameter optimization and subband reconstruction. Firstly, the Q factor of RSSD and the number of decomposition layers were adaptively selected using the artificial fish swarm algorithm to construct the optimal wavelet basis matching the fault features and to obtain the low resonance components containing transient components. Secondly, the optimum sub-band which carried transient feature information, was selected and reconstructed using the proposed subband screening principle. Finally, the periodic impulses of the composite fault signals were identified and extracted by MOMEDA method. The analysis on the simulated signals and the experimental composite fault signals in the bearing life cycle shows that the proposed method may effectively extract each fault feature from the composite fault signals, and accurately realize the composite fault diagnosis compared with RSSD-maximum correlation kurtosis deconvolution(RSSD-MCKD) method. © 2022 China Mechanical Engineering Magazine Office. All rights reserved.