基于归档集特性的自适应多目标粒子群优化算法  

Adaptive multi-objective particle swarm optimization algorithm based on archive properties

作　　者：王强[1] 郑力明[1] 

机构地区：[1]暨南大学电子工程系,广州510000

出　　处：《计算机应用研究》2015年第11期3291-3294,共4页Application Research of Computers

基　　金：广东省教育部产学研结合重点项目(2011A090200085);广东省战略性新兴产业发展专项资金资助项目(2011912010);广东省教育部产学研结合项目(2012B091000046)

摘　　要：为了改善多目标粒子群算法收敛性能及跳出局部最优的能力,通过研究算法运行过程中归档集规模的变化,总结归档集规模的变化规律,并依此规律提出一种近似确定算法运行状态的简单方法。基于算法运行状态的近似确定,可以自适应调整参数。实验分析与比较结果显示,自适应方法能够有效地加速收敛,而且对于一些复杂问题有较好的跳出局部最优的能力。实验结果同时表明,近似确定算法状态的方法对于参数的调整是简单且有效的。In order to improve the convergence performance and the ability to jump out of local optimum of the multi-objective particle swarm optimal algorithms, this paper summarized the laws for the archive size development, and proposed a simple method which be used to determine the status of the algorithm. Based on this method, the parameters could be adjusted adaptively. The experimental analysis show that proposed method has good convergence performance and has better capacity of jump out of local optimum. The experiments results prove that the method of determine the status of the algorithm is simple and effectively.

关 键 词：多目标粒子群算法 归档集规模 算法运行状态 参数自适应 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]