Optimal Preventive Maintenance for Systems Based on Copula and Survival Signature Under Random Inspection

Abstract

Under random inspection, some partial system maintenance models still do not account for dependence among components, which may bias system reliability assessment and consequently affect the optimality of maintenance strategies. To address this issue, this paper proposes an optimal preventive maintenance (PM) strategy for multi‐type component systems with component dependence under random inspection. The proposed strategy employs copula theory to characterize dependence among component lifetimes and combines it with the survival signature to calculate probability terms related to system reliability, thereby deriving the expected cost and the mean length of the system under four mutually exclusive maintenance scenarios. On this basis, the mean cost per unit time of the system is obtained, and the optimal PM strategies under different dependence conditions can be determined by jointly optimizing the scheduled PM time and the PM triggering threshold. A real‐data case study of a wind turbine generator system verifies the optimality and applicability of the proposed strategy.