Multi-objective optimization using genetic algorithm: Applications to imperfect preventive maintenance model

In this study, a multi-objective hybrid genetic algorithm (MOHGA) is proposed to optimize a multi-objective imperfect preventive maintenance (MOIPM) model. The MOHGA proposed not only utilizes a Pareto-based technique to determine and retain the superior chromosomes as the GA chromosome evolutions are performed, but also guides their search direction. In order to obtain diverse non-dominated solutions that approach the optimized Pareto-efficient frontier, the closeness metric and diversity metric are employed to evaluate the superiority of the non-dominated solutions. Accordingly, decision makers can easily determine the most appropriate maintenance alternative to constitute a maintenance strategy from the optimized non-dominated solutions, given the practical requirements of system performance under the constraints of maintenance resources. Furthermore, this study employs response surface methodology via systematic parameter experiments to determine the best search parameter settings in the MOHGA proposed. A simulated case verifies the efficacy and practicality of the MOHGA.