Design optimization of multi-choice subsystems under mixed failure dependency models using computational intelligence
Abstract
The design and optimization of systems with multiple subsystems under mixed failure dependencies and multi-choice redundancy are critical in enhancing system performance. This paper addresses this challenge by proposing a multi-choice subsystem design optimization approach that minimizes system cost while maintaining the system’s availability within a predefined limit. Three nature-inspired optimization algorithms are employed: differential evolution (DE), Coulomb’s and Franklin’s laws algorithm (CFA), and grey wolf optimizer (GWO) to solve the problem. The proposed approach is tested on a series–parallel system with ten subsystems, where mixed failure dependencies and redundancy choices are considered. Results show that the CFA outperforms both DE and GWO in terms of system cost, function evaluations (NFE), and CPU time. Additionally, the CFA demonstrated lower standard deviation, indicating more consistent and reliable performance. A comprehensive comparison of the algorithms, including system cost, NFE, CPU time, and SD, is presented. The best solution obtained by the CFA includes the selection of redundancies, repair teams, and choices, which ensures system availability while minimizing costs. This study highlights the effectiveness of the proposed approach in handling system design problems involving mixed failure dependencies and multi-choice subsystems.