Fault Diagnosis of Rolling Bearing Based on Improved Data Fusion

In view of the high misdiagnosis and low reliability of single diagnosis method for rolling element bearings, a hybrid diagnosis algorithm based on variational modal decomposition (VMD)-support vector machine (SVM) and mathematical morphology (MM)-correlation analysis (CA) was proposed. Parallel dual-channel diagnosis was adopted in the method. In the first channel, VMD was used to decompose fault signals in the frequency domain to obtain signal characteristics, and then Bayesian SVM classifier was employed to obtain the posterior probability of diagnosis results, which has the advantage of high diagnosis precision. In the second channel, MM method was used to extract fault features in the time domain, to obtain the correlation coefficient of diagnosis results through CA method, which has strong generalization ability. Subsequently, the improved weighted average evidence theory combined the two-channel decision results, and made full use of the advantages of the two single methods to complement each other, so as to realize fault diagnosis in hybrid method. Finally, the new method was validated by bearing fault test bench data and compared with the single method. The results show that the method can effectively extract fault features from non-stationary signals and improve the reliability of the diagnosis. © 2022, Department of Journal of the China Railway Society. All right reserved.