Multiobjective simulation optimization using an enhanced genetic algorithm

This paper presents an improved genetic algorithm approach, based on new ranking strategy, to conduct multiobjective optimization of simulation modeling problems. This approach integrates a simulation model with stochastic nondomination-based multiobjective optimization technique and genetic algorithms. New genetic operators are introduced to enhance the algorithm performance of finding Pareto optimal solutions and its efficiency in terms of computational effort. An elitism operator is employed to ensure the propagation of the Pareto optimal set, and a dynamic expansion operator to increase the population size. An importation operator is adapted to explore some new regions of the search space. Moreover, new concepts of stochastic and significant dominance are introduced to improve the definition of dominance in stochastic environments.