| 摘要: |
| 在求解多目标优化问题时,微粒群优化算法有容易陷于局部极值、函数评价次数多和受到维数限制等不足之处.提出了一种基于合作型协同和ε-占优的多目标微粒群算法(cooperative coevolutionary and ε-dominancebased multi-objective particle swarm optimizer,简称CEPSO).依据决策变量分解问题,采用多个子群分别优化各个子问题,并在更新粒子位置时采用均匀分布变异算子防止微粒群早熟收敛;在保存非劣解时,使用< |
| 关键词: 多目标优化 进化算法 合作型协同进化 ε-占优 多目标微粒群算法 |
| DOI： |
| 分类号: |
| 基金项目:Supported by the National Natural Science Foundation of China under Grant No.60374054 (国家自然科学基金); the Natural Science Foundation of Shandong Province of China under Grant Nos.Y2003G01, Z2004G02 (山东省自然科学基金) |
|
| A Cooperative Coevolutionary and ε-Dominance Based MOPSO |
|
ZHENG Xiang-Wei,LIU Hong
|
| Abstract: |
| Particle Swarm Optimizers (PSOs) have been applied to solve Multi-Objective Optimization Problems (MOPs) for its successful applications in solving single objective optimization problems and are named as Multi-Objective PSOs (MOPSOs). However, MOPSOs are often trapped in local optima, cost more function evaluations and suffer from the curse of dimensionality. A cooperative coevolutionary and ε-dominance based MOPSO (CEPSO) is proposed to attack the above disadvantages. In CEPSO, the MOPs are decomposed according to their decision variables and are optimized by corresponding subswarms respectively. Uniform distribution mutation operator is adopted to avoid premature convergence. All subswarms share one archive based on ε-dominance, which is also used as leader set. Collaborators are selected randomly from archive and used to construct context vector in order to evaluate particles in subswarm. CEPSO is tested on several classical MOP benchmark functions and the simulation results show that CEPSO can escape from local optima, optimize high dimension problems and generate more Pareto solutions. Therefore, CEPSO is competitive in solving MOPs. |
| Key words: multi-objective optimization evolutionary algorithm cooperative coevolution ε-dominance MOPSO (multi-objective particle swarm optimizer) |
