A hybrid intelligent algorithm for reliability optimization problems

This paper focuses on general cold standby redundancy systems with imperfect switches and the lifetime of system is modeled as a fuzzy variable. The system performance - α-system lifetime - characterized in the context of credibility is investigated. In order to estimate the system performance, a fuzzy simulation is designed. A standby redundancy fuzzy chance-constrained programming model is established to optimize this system performance under cost constraint. In order to solve the proposed model, we also design a hybrid intelligent algorithm which uses fuzzy simulation to generate a training data set, the back propagation algorithm to train a neural network to approximate the uncertain function and genetic algorithm to optimize the system performance. Finally, a numerical experiment is discussed to illustrate the idea of the modeling and the effectiveness of the proposed algorithm.