Locate multiple pareto optima using a species-based multi-objective genetic algorithm

Abstract

In many real-world multi-objective optimization problems (MOOPs), the decision maker may only concern partial, rather than all, Pareto optima. This requires the solution algorithm to search and locate multiple Pareto optimal solutions simultaneously with higher accuracy and faster speed. To address this requirement, a species-based multiobjective GA (speMOGA) is designed in this paper, where multiple subpopulations would be constructed to search for multiple nondomiated solutions in parallel via decomposing a MOOP into a set of subproblems using the Tchebycheff approach. Based on a series of benchmark test problems, experiments are carried out to investigate the performance of the proposed algorithm in comparison with two classical multi-objective GAs: MOEA/D and NSGA-II. The experimental results show the validity of the proposed algorithm on locating multiple Pareto optima.