Particle filter-based approach to estimate remaining useful life for predictive maintenance

Abstract

Estimation of remaining useful life (RUL) plays a vital role in performing predictive maintenance for complex systems today. However, it still remains a challenge. To address this issue, we propose a Particle filter (PF)- based method to estimate remaining useful life for predictive maintenance by employing PF technique to update the nonlinear predictive models for forecasting system states. In particular, we applied PF techniques to estimate remaining useful life by integrating data-driven modeling techniques in order to effectively perform predictive maintenance. After introducing the PF-based algorithm, the paper presents the implementation along with the experimental results through a case study of Auxiliary Power Unit (APU) starter prognostics. The results demonstrated that the developed method is useful for estimating RUL for predictive maintenance.