An efficient hybrid multi-objective particle swarm optimization with a multi-objective dichotomy line search

JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS

卷:280

Article

Xu, Gang¹  Yang, Yu-qun²  Liu, Bin-Bin¹  Xu, Yi-hong¹  Wu, Ai-jun¹ 

[1] Nanchang Univ, Dept Math, Nanchang 330031, Jiang Xi, Peoples R China

[2] Nanchang Univ, Middle Sch, Nanchang 330047, Jiang Xi, Peoples R China

关键词: Particle swarm optimization;    Multi-objective optimization;    Multi-objective dichotomy linear search;    Non-dominated solutions;   

DOI  :  10.1016/j.cam.2014.11.056

来源: Elsevier

【 摘 要 】

Recently more research works are focused on multi-objective particle swarm optimization algorithm (MOPSO) due to its ability of global and local search for solving multi-objective optimization problems (MOOPs); however, most of existing MOPSOs cannot achieve satisfactory results in solution quality. This paper proposes an efficient hybrid multi-objective particle swarm optimization with a multi-objective dichotomy line search (MOIS), named MOLS-MOPSO, to deal with such problem. MOLS-MOPSO combines an effective particle updating strategy with the local search of MOLS. The effective particle updating strategy is used for global search to deal with premature convergence and diversity maintenance within the swarm; the MOLS is periodically activated for fast local search to converge toward the Pareto front. The exploratory capabilities are enhanced more efficiently by keeping a desirable balance between global search and local search, so as to ensure sufficient diversity and well distribution amongst the solutions of the non-dominated fronts, while retaining at the same time the convergence to the Pareto-optimal front. Comparing MOLSMOPSO with various state-of-the-art multi-objective optimization algorithms developed recently, the comparative study shows the effectiveness of MOLS-MOPSO, which not only assures a better convergence to the Pareto frontier but also illustrates a good diversity and distribution of solutions. (C) 2014 Elsevier B.V. All rights reserved.

【 授权许可】

Free