2304 パレート適合度を用いた多目的PSO(Particle Swarm Optimization)(OS8-1 近似最適化I,OS8 設計と最適化)

Abstract

In this paper, we propose a novel multi-objective particle swarm optimization (MOPSO) algorithm considering the diversity of the solutions both in the objective space and decision variable space. There was a problem that developed algorithm in the laboratory of us before consider the diversity of the solutions only in the objective space. Then, it was necessary to consider both in the objective space and decision variable space. Fitness function is a concept that can classify solutions into non-inferior solutions and inferior solutions in the decision variable space. We propose new mechanism to find the pareto optimal solution in the decision variable space using it. We use fitness function to select the g-best among the inferior solutions in the decision variable space. Moreover, we improved update method of p-best. The validity of proposed approach is examined through some typical numerical examples. In addition, we compare proposed algorithm to developed algorithm in the laboratory of us before.