基于种群年龄模型的动态粒子数微粒群优化算法  

Dynamic particle population particle swarm optimization based on population age model

作　　者：江善和[1,2] 纪志成[1] 张日东[3] 沈艳霞[1] 

机构地区：[1]江南大学电气自动化研究所,无锡214122 [2]安庆师范学院物理与电气工程学院,安庆246011 [3]杭州电子科技大学信息与控制学院,杭州310018

出　　处：《系统工程理论与实践》2012年第11期2550-2556,共7页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(61174032;61273101);教育部博士点基金(200802950004);安徽高校省级自然科学重点项目(KJ2011Z232)

摘　　要：针对微粒群优化算法的早熟停滞缺陷问题,提出了一种基于种群年龄模型的动态粒子数微粒群优化算法.该算法建立了生物种群年龄模型,将每个粒子划分为不同的年龄段.动态地依据种群环境和个体信息有效地控制种群的粒子数规模;设计了较优粒子的生殖策略和较差粒子的死亡策略,增加群体的多样性和减少冗余计算量.以保证算法获得最优性能.将此算法与其他改进算法进行比较,仿真测试结果表明,新算法具有较高的全局搜索成功率和效率.计算量显著降低.优化精度显著提高,能够有效地避免算法陷入局部停滞的缺点.For the problem that particle swarm optimization （PSO） algorithm often suffers from being trapped in local optima so as to be premature convergence, a dynamic particle population PSO based on population age model is proposed to efficiently control premature stagnation. Firstly, life population age model is constructed, which divides diverse age group for a certain particle, and effectively regulates population size in accordance with population environment and individual information. Secondly, the particle reproduction strategy of the good particle for keeping the diversity of swarm and the particle vanishing strategy of the worst particle for decreasing optimal performance is guaranteed in this algorithm. excessive calculated amount are designed, so the Finally, the comparison experiments have been made with four benchmark functions between the proposed algorithm and other improved PSO. The simulation experimental results show that the proposed method not only greatly improves the global successful searching probability and searching efficiency, but also effectively avoids the local stagnation problem.

关 键 词：微粒群优化算法 早熟停滞 种群年龄模型 动态粒子数 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]