A new cost evaluation model of predictive maintenance for dynamic systems with hidden degradation

This paper proposes a predictive maintenance policy for a dynamic system suffering from hidden degradation. Particle filtering is adopted to estimate and predict system state. With the knowledge of system state, the predictive maintenance policy is adopted. In order to evaluate the performance of the proposed maintenance policy, a mathematical model is set up with the goal of minimizing the long-term expected cost rate. Different from previous studies, this paper incorporates the degradation-caused cost as one of the cost items. This paper illustrates that the inspection interval is associated with the system reliability. And the main parameters that need to be determined are the thresholds for the system reliability and the preventative maintenance (PM) strategy, Monte Carlo strategy is used to compute the optimal thresholds. For comparison, we also simulate the long-term expected cost without the degradation-caused cost, and the result shows the effectiveness of the proposed model.