摘要
提出了一种改进型多目标粒子群优化算法(MOPSO-Ⅱ).该算法为粒子群中每个粒子增加一个"扰动向量",以利于粒子跳出局部最优并为粒子的全局最优位置赋予了时限的属性,可防止过于频繁地更新全局最优位置,有利于增强粒子搜索的持效性.该算法改进了粒子越界的处理方法,最大程度上保持粒子优秀的搜索方向.通过典型的多目标测试函数ZDT对该算法进行测试,实验结果表明,带ε-支配的MOPSO-Ⅱ算法在解群的分布性方面要优于使用了拥挤距离机制MOPSO-Ⅱ算法和NSGA2算法,对比实验还表明MOPSO-Ⅱ算法在收敛性方面要优于NSGA2.因此,MOPSO-Ⅱ在求解多目标优化问题上有一定优势,是一种有前途的算法.
This paper proposed an improved multi-objective particle swarm optimization algorithm（MOPSO-Ⅱfor short）.In this algorithm a disturbance vector was added to each particle to enhance the ability of escaping from local optima.Secondly,aproperty of limited time was assigned to each global best particle to keep effective sustainable search.Finally,an improved boundary treatment method was proposed to preserve the ercellent search direction.Some experiments on a set of classical benchmark functions were conducted in the paper.Experimental results show that MOPSO-Ⅱ usingε-dominance mechanism performed better than those using crowding distance method of MOPSO-Ⅱand NSGA2on the distribution of solutions.It also shows that the proposed algorithm has better convergence than NSGA2.Consequently,MOPSO-Ⅱhas some advantages in the field of solving multi-objective optimization problems and it will be a promising algorithm.
出处
《武汉大学学报（理学版）》
CAS
CSCD
北大核心
2014年第2期144-150,共7页
Journal of Wuhan University:Natural Science Edition
基金
国家自然科学基金(61165004
61305079)
江西省自然科学基金(20114BAB201025)
福建省自然科学基金(2012J01248)
江西省教育厅科技项目(GJJ12307
GJJ13320)
福建省杰出青年培育计划(福建省教育厅[2011]29号)
河北省教育厅科研项目(QN20131053)
河北省科技计划项目(13210331)
河北省青年拔尖人才支持计划(冀字[2013]17号)
关键词
粒子群优化
多目标优化
扰动向量
particle swarm optimization
multi-objective optimization
disturbance vector
