Computational Optimization of Preventive Maintenance Schedules in Repairable Mechanical Systems Using NSGA-II

In this study, a computational framework is proposed for optimizing preventive maintenance scheduling in complex mechanical systems, with a focus on minimizing total maintenance costs while preserving system availability and mechanical reliability. The model incorporates multi-level maintenance actions—including inspection, repair, and component replacement—over a defined planning horizon. A nonlinear integer programming formulation is developed to capture cost elements such as random failure, repair, replacement, and planned downtime. To address the combinatorial complexity of the problem, a Non-Dominated Sorting Genetic Algorithm II (NSGA-II) is employed to generate near-optimal solutions. The proposed method is applied to a real-world Cathodic Protection System used in steel gas distribution networks, which are critical mechanical infrastructures subject to electrochemical degradation. Results demonstrate a 36% reduction in total maintenance costs, highlighting the effectiveness of the model in improving asset performance and extending system life through optimized maintenance strategies.