Author = Mustafa Zandieh

Optimization of Redundancy Allocation Problem with Non-exponential Repairable Components Using Simulation Method and Artificial Neural Networks

Volume 9, Issue 3, Autumn 2019, Pages 226-243

Maghsoud Amiri, Mojtaba Hemmati, Mustafa Zandieh

Abstract In this study, a new bi-objective model along with a novel solving method are provided to address the non-exponential RAP in series-parallel systems with repairable components. The proposed method is based on optimization via simulation approach and artificial neural network technique. In addition, to be more realistic, discounts strategies for purchasing the components are employed during modeling. The main objective of the model is to maximize mean time to the first failure (MTTFF) of the system via allocating the best redundant components for each subsystem. Since, the components’ failure rate has non-exponential distribution, the simulation technique and ANN are applied to find the MTTFF. To solve the problem, some meta-heuristic algorithms integrated with the simulation method. Several numerical examples are carried out to test the proposed approach and as the results show, the proposed approach is much more real than previous ones and also the near optimum solutions are achieved.