Abstract: In k-out-of-n systems, degradation dependency, failure dependency, and economic dependency are commonly present among units, and it is difficult to effectively formulate maintenance strategies based on the assumption of unit independence or a single type of dependency. Therefore, considering the multiple dependencies in k-out-of-n systems, research on condition-based system maintenance strategy optimization is conducted. Initially, the Wiener process is employed to model the degradation performance of individual units, and a k-out-of-n system degradation model that incorporates both degradation and failure dependencies is proposed through factor analysis and a load-averaging allocation mechanism, based on which the system reliability function with arbitrary initial degradation levels is derived. Subsequently, to address the economic dependency arising from shared setup costs among units, a joint optimization model for preventive maintenance and opportunistic maintenance is established with the objective of minimizing long-term maintenance cost rates. Finally, the steady-state probability of system degradation is calculated using semi-regenerative processes, and the joint optimization model is solved by a genetic algorithm. Consequently, optimal thresholds for preventive maintenance and opportunistic maintenance as well as inspection intervals are identified to realize the formulation of maintenance strategies for k-out-of-n systems with degradation dependency, failure dependency, and economic dependency. The numerical study demonstrates that the proposed maintenance optimization method can effectively support the operation and maintenance management of complex engineering systems.

Key words: maintenance optimization, k-out-of-n system, degradation dependency, failure dependency, economic dependency