基于混沌知识迁移的的多种群粒子群文化算法  

Multi-population particle swarm cultural algorithms adopting chaotic knowledge migration

作　　者：郭一楠[1] 程健 曹媛媛[1] 刘丹丹[1] 

机构地区：[1]中国矿业大学信息与电气工程学院,江苏徐州221116

出　　处：《控制理论与应用》2011年第9期1056-1062,共7页Control Theory & Applications

基　　金：国家自然科学基金资助项目(60805025);江苏省自然科学基金资助项目(BK2010183);江苏省青蓝工程资助项目(2008)

摘　　要：在已有的多种群粒子群文化算法知识迁移策略中,迁移知识不一定能反映优势区域中的较优点.为提高知识迁移效率,在知识迁移机制中引入混沌搜索策略,提出一种多种群粒子群文化算法的混沌知识迁移策略.它利用混沌序列对迁移单元进行深入探索,以提高迁移知识的有效性;根据进化代数动态调整知识迁移间隔,从而在进化前期维持种群的多样性,在进化后期加速种群收敛.数值计算结果表明,该算法可以有效提高进化收敛速度,帮助子种群跳出局部较优解.In existing multi-population particle swarm cultural algorithms based on knowledge migration,the migrated knowledge may not reflect the best individuals in the advantageous region.In order to improve the efficiency of knowledge migration,a novel multi-population particle swarm cultural algorithms adopting chaotic knowledge migration is proposed.Chaos sequence is used to deeply explore migrated cells for improving the validity of migrated knowledge.Knowledge migration interval is then dynamically adjusted by generation.This keeps the diversity of population in the early evolution and accelerates the convergence in the latter evolution.Simulation results indicate that the algorithm effectively improves the speed of convergence and eliminates premature convergence.

关 键 词：混沌序列 知识迁移 多种群 粒子群文化算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]