A predictive opportunistic maintenance policy for a serial–parallel multi-station manufacturing system with heterogeneous components

This paper proposes a predictive opportunistic maintenance (OM) policy tailored for a serial–parallel multi-station manufacturing system (SP-MMS) characterized by heterogeneous degradation of critical components, alongside economic and structural dependencies among them. The OM policy comprises interactive component-level predictive maintenance (PdM) scheduling and system-level OM optimization. In the component-level PdM scheduling, online degradation data for each critical component are utilized to dynamically estimate the posterior distribution of its degradation rate and its remaining useful life (RUL), ensuring precise RUL predictions. Leveraging this real-time RUL prediction, the optimal preventive maintenance (PM) time for each critical component is determined based on a cost rate model. System-level OM optimization is initiated whenever the optimal PM time for a critical component is reached, aiming to identify the optimal group of components for OM. A novel group-cost-saving indicator, consisting of penalty cost, saving of production loss and saving of maintenance-setup cost, is proposed to determine the optimal OM group, considering the two dependencies. The OM optimization is formulated as a mixed integer linear programming problem, and component classification based on the structural dependence is proposed to reduce the solution space. Two case studies are conducted to validate the effectiveness and applicability of the proposed OM policy.