An improved Wiener process model with adaptive drift and diffusion for online remaining useful life prediction

Remaining useful life (RUL) prediction plays an important role in the field of prognostics and health management (PHM). Although several Wiener process models with adaptive drift have been developed for RUL prediction, these models assume the diffusion parameter is fixed and therefore fail to capture the real degradation process. Accordingly, this paper proposes an improved Wiener process model for RUL prediction, in which both drift and diffusion parameters are adaptive with the updating of monitoring data. The proposed model considers the quantitative relationship between degradation rate and degradation variation. When a new monitoring data is available, we update the model parameters and therefore the RUL distribution by applying recursive filter and expectation maximization (EM) algorithm. In addition, a prediction region is constructed based on the 3σ-interval criterion to eliminate the abnormal monitoring data, followed by a model selection method developed to compare the prediction accuracy of the proposed model with the existing models. The proposed model's superiority and the effectiveness of the model selection method are illustrated and validated by an application to the identical thrust ball bearings.