Remaining life prediction of rolling bearings based on the WV-EWMA and the Wiener process with random failure thresholds

ObjectiveAiming at the characteristics of stable stage and degradation stage during the performance evolution process of bearings, this study realizes the prediction of the remaining useful life of rolling bearings.MethodsFirstly, considering that the random distribution of performance parameters in the stable stage is sometimes unknown, a bearing health condition monitoring and change point detection method based on the weighted variance-exponentially weighted moving average (WV-EWMA) control chart was proposed to identify the initial degradation time. Secondly, for the degradation stage, a degradation characterization model based on the Wiener process was established. Taking into account the individual differences of bearings, the failure threshold was assumed to be a random variable, and the remaining useful life distribution model of bearings was derived. Thirdly, the remaining useful life of rolling bearings were predicted online using the maximum likelihood estimation method and Bayesian theory. Finally, engineering verification was conducted via tests of type 16004 rolling bearings.ResultsThe results indicate that, compared to traditional methods, the prediction accuracy for the remaining useful life of rolling bearing of the proposed method is improved by at least 55.98%.