Optimal predictive maintenance policy for multi-state deteriorating system under periodic inspections

This paper proposes a predictive maintenance model for multi-state Markov deteriorating system. In this model, a system subject to aging and fatal shocks and with states 0(initial condition)<1<2<...<L(failure condition) is described. Then, a control limit maintenance policy is proposed where the system is inspected periodical with period T to identify the system state and the action from {do-nothing, repair, replacement} is taken after that. The policy considers three maintenance variables (preventive maintenance threshold, failure threshold and inspection interval) to evaluate the long-time system availability. Furthermore, algorithms are developed to implement the optimal policies on Markov deteriorating system. Finally, the effectiveness and efficiency of the optimal policies is illustrated by numerical examples.