Multi-objective optimization in reliability system using genetic algorithm and neural network

To optimize the design of reliability systems, an analyst is frequently faced with the demand of achieving several targets (i.e., maximization of system reliability, minimizations of cost, volume, and weight), some of which may be in conflict with each other. This paper presents a novel hybrid approach, combining a multi-objective genetic algorithm and a neural network, for multi-objective optimization of a reliability system, namely GANNRS (Genetic Algorithm and Neural Network for Reliability System optimization). The multi-objective genetic algorithm's evolutionary strategy is based on the modified neighborhood design, and is presented to find the Pareto optimal solutions so as to provide a variety of compromise solutions to the decision makers. The purpose of the neural network is to generate a good initial population in order to speed up the searching by genetic algorithm. For demonstrating the feasibility of the proposed approach, four multi-objective optimization problems of reliability system are used, and the outcomes are compared with those from other methods. The evidence shows that the proposed GANNRS is more efficient in computation, and the results from the objectives are appealing.