Solving hard multiobjective problems with a hybridized method

Abstract

This paper presents a hybrid method to solve hard multi- objective problems. The proposed approach adopts an epsilon-constraint method which uses a Particle Swarm Optimizer to get points near of the true Pareto front. In this approach, only few points will be generated and then, new intermediate points will be calculated using an interpola- tion method, to increase the among of points in the output Pareto front. The proposed approach is validated using two difficult multiobjective test problems and the results are compared with those obtained by a multiobjective evolutionary algorithm representative of the state of the art: NSGA-II.