一种新的求解动态连续优化的分层粒子群算法  

A new hierarchical PSO algorithm for solving dynamic and continuous optimization problems

作　　者：朱庆保[1,2] 徐晓晴[1,2] 朱世娟[3] 

机构地区：[1]南京师范大学计算机科学与技术学院,南京210097 [2]南京师范大学江苏省信息安全保密技术工程中心,南京210097 [3]安庆师范学院计算机信息学院,安徽安庆246001

出　　处：《控制与决策》2013年第10期1573-1577,共5页Control and Decision

基　　金：国家自然科学基金项目(60673102;61073118);江苏省高校自然科学基金项目(10KJD520004)

摘　　要：为了高效求解动态连续优化问题,提出一种分层粒子群优化算法.该算法将动态函数定义域分成个子空间,每个空间用一个粒子群作为第一层进行独立搜索,个子空间的最优粒子再组成一个全局粒子群进行全局搜索,以达到全局牵引的作用,同时提出探测环境和响应环境的策略.利用经典的动态函数对算法进行测试,结果表明所提出算法能够迅速适应环境变化和跟踪最优解的变化,效果令人满意.In order to efficiently solve the dynamic and continuous optimization problems, a hierarchical PSO algorithm is proposed. The solution space of a dynamic optimization probtem is divided into Q subspaces. One particle swarm is assigned to each subspace to search independently as the first layer. Then the best particles in the Q particle swarms as the second layer search the global domain so as to guide the other particles in each subspace. Moreover, an environment detection and response strategy is proposed. Numerical experiments on the classical dynamic function demonstrate that the algorithm can adapt dynamic environment and track a changing optimum quickly.

关 键 词：粒子群算法 动态优化 连续优化 分层 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]