Correlation‐driven multivariate degradation modeling and RUL prediction based on Wiener process model

What is it about?

For rotary machinery systems, shock pulse-based degradation data contains sufficient health information and is widely used for remaining useful life (RUL) prediction. The rapid development in engineering and online monitoring technologies changes the degradation data from one performance characteristic (PC) to multiple PCs. Due to the common environmental condition, a strong correlation in degradation paths for multiple PCs is generally observed. In this paper, we model the actual degradation processes of multiple PCs with the multiple Wiener processes, and the degradation correlation is incorporated by the commonly shared environmental condition function. Thus, it directly catches the strong correlation of degradation rates and volatilities to all dimensions of multiple PCs. Our model is more economic in the number of parameters, and is more flexible because it can apply to both linear and nonlinear degradation process. The correlation coefficients and RUL distribution approximation are provided in closed-forms. By the same token, we extend the proposed model to two-stage degradation process so that multiple PCs are correlated in each stage. An adaptive drift is adopted in the Wiener process model for the real-time degradation states update. With the arrival of new monitoring data, we update the model parameters with the help of the Bayesian algorithm and the expectation maximization (EM) algorithm. The proposed model’s effectiveness is illustrated by numerical studies and a real-world application to the wheel treads on high-speed trains.

Featured Image

Photo by Daniel Abadia on Unsplash

Photo by Daniel Abadia on Unsplash