Remaining useful life prediction for nonlinear degrading systems with maintenance

Abstract

Remaining useful life (RUL) prediction is one of the most critical procedures of the prognostics and health management (PHM). In the existing literature, most RUL prediction methods are under the assumption that there is no maintenance activity during the whole life time of the degrading system. However, most practical systems experience various kinds of maintenance activities when they are in operation. This article presents an approach to predict the RUL of a class of nonlinear degrading systems with stochastic maintenance. To predict the RUL for systems with stochastic maintenance, a wiener process based degradation model is proposed. The switches between states of normal operation and maintenance are described by a continuous time Markov chain (CTMC). In addition, the maximum likelihood estimation (MLE) is adopted to estimate both unknown parameters in the degradation model and the transition probability between normal operation and maintenance. The analytical form of first hitting time (FHT) of degradation process is difficult to derive with the presence of maintenance activities. To avoid complicated mathematical derivation of stochastic differential, Monte Carlo method is used to obtain a numerical result of the RUL distribution. A numerical study is presented to illustrate and validate the proposed method.