Two-Phase Degradation Modeling and Residual Life Prediction Based on Nonlinear Wiener Process

Abstract

Aiming at the problem of inaccurate description of two-phase degradation of products, a two-phase nonlinear Wiener model was established, and a residual life prediction method was proposed. Firstly, considering the random effect of the degenerate change-point, a two-phase nonlinear Wiener degradation model is established by using the normal distribution to describe the drift parameters of each phase. Secondly, based on Bayesian theory, the posteriori distribution of model parameters is derived, and the MHGS method is proposed to estimate the parameters of the two-phase degradation model. Then, a method to determine the degradation stage was proposed, based on DIC criterion. Combined with the state-space model and Kalman filter, the online updating process and residual life probability distribution of the two-phase degradation model were deduced. Finally, the proposed model and method are validated by solder joint degradation data. The results show that the proposed method can accurately estimate the model parameters and predict the residual life. Compared with the two-phase model of linear Wiener process, the two-phase model of nonlinear Wiener process proposed in this paper has higher prediction accuracy.