Residual life prediction of wind turbine bearings based on multivariate Wiener process

Regarding the issue of accelerated performance degradation of wind turbine bearings due to various factors affecting their performance, a prediction method of bearing residual life based on ternary Wiener process is proposed. Firstly, two bearing vibration signals in perpendicular directions to each other and bearing temperature signals are analyzed, and the performance indexes representing the bearing health state are constructed respectively. Secondly, according to the characteristics of bearing performance degradation process, a ternary degradation model based on Wiener process is established. Then, the appropriate Copula function is selected by using AIC (Akaike Information Criterion) information criterion to analyze the correlation characteristics between the two performance indicators, and Vine-Copula function is used to correlate the multiple Copula functions. The joint probability density function of bearing remaining life is calculated. The distributed maximum likelihood estimation method is used to update the model parameters online to predict the change of bearing health state at future time and obtain its remaining life. Finally, the proposed method is validated and analyzed using actual bearing degradation data, and the results showed that it can effectively predict the remaining life of bearings. Compared with the prediction methods based on one-dimensional and two-dimensional Wiener processes, the proposed method has better prediction accuracy.