Optimal predictive maintenance for a nonstationary gamma process

Nonstationary gamma processes have extensive applications in depicting the degradation of many practical systems. This paper proposes a predictive maintenance policy that involves various types of maintenance actions for a nonstationary gamma process. Periodic inspections fully reveal the degradation levels of the system. The information on age, degradation, and the number of conducted preventive maintenance actions is synthesized for decision-making, which distinguishes our model from most existing models considering only degradation states. The objective is to find the maximum number of repairs and the best thresholds for preventive maintenance by minimizing the expected average cost in an infinite time horizon. The maintenance problem is addressed as a semi-Markov decision problem. An optimization algorithm is developed to find the optimal values of the decision variables. The effectiveness of the proposed method is verified by a coating system.