Remaining useful life prediction for multi-component systems with stochastic correlation based on auxiliary particle filter

Abstract

The remaining useful life (RUL) prediction of a complex system requires accurate evaluation of component degradation states and a full understanding of how these states are expected to evolve. These challenges become more complicated when stochastic correlations exist between components. To address this issue, a nonlinear Wiener process degradation model is proposed, which comprehensively considers the inherent degradation of a component and the influence of related components’ degradation levels. The degradation process of each component is modeled as a nonlinear Wiener process, and the deterioration induced by other components is described by a nonlinear function. Subsequently, an online RUL prediction method is developed for multi-component systems with varying structures. Implicit degradation states and unknown parameters are jointly estimated using auxiliary particle filtering (APF) and maximum likelihood estimation (MLE) algorithms and updated in real time according to observed data. Finally, the effectiveness and practicality of the proposed method is verified through a numerical simulation experiment and case studies of an aircraft turbine engine and a gearbox system.