A New Multiobjective Particle Swarm Optimizer with Fuzzy Learning Sub-Swarms and Self-Adaptive Parameters

摘要

Multiobjective optimization (MOO) problem are usually very computationally expensive since there are usually exponentially large Pareto-optimal solutions. The paper presents a new multiobjective particle swarm optimization (MOPSO) algorithm which uses fuzzy learning sub-swarms and self-adaptive parameters to improve the overall search ability. During the search process, each particle in the swarm can have a sub-swarm of p particles which are sequentially generated based on fuzzy-controlled parameters, and a fuzzy satisfying solution is chosen to replace the particle in the next generation of the swarm. Numerical experiments and case studies demonstrate that our approach can achieve good solution qualities with low computational costs.