Improved rank-niche evolution strategy algorithm for constrained multiobjective optimization

Abstract

Purpose - The purpose of this paper is to improve and to extend the use of original rank-niche evolution strategy (RNES) algorithm to solve constrained and unconstrained multiobjective optimization problems. Design/methodology/approach - A new mutation step size is developed for evolution strategy. A mixed ranking procedure is used to improve the quality of the fitness function. A self-adaptive sharing radius is developed to save computational time. Four constraint-treating methods are developed to solve constrained optimization problems. Two of them do not use penalty function approach. Findings - The improved RNES algorithm finds better quality Pareto-optimal solutions more efficiently than the previous version. For most test problems, the solutions obtained by improved RNES are better than, or at least can be compared with, results from other papers. Research limitations/implications - The application of any evolutionary algorithm to real structural optimization problems would face a problem of spending huge computational time. Some approximate analysis method needs to be incorporated with RNES to solve practical problems. Originality/value - This paper provides an easier approach to find Pareto-optimal solutions using an evolutionary algorithm. The algorithm can be used to solve both unconstrained and constrained problems.