摘要
传统PSO算法的收敛性能会随求解问题空间维数的增加而下降,根据协同进化原理,通过对传统PSO算法进行协同优化处理,设计一种改进的协同PSO算法。在每次迭代过程中,采用传统PSO算法更新粒子位置和速度,由此产生群体当前的全局最优位置;对所有粒子所经历的最优位置进行降维的协同优化,产生一个参考全局最优解;根据参考全局最优解更新群体当前的全局最优位置。仿真结果表明,该算法可以明显提高收敛速度,在某些问题上可以收敛到全局最优。
Traditional PSO algorithms suffer from the curse of dimensionality which implies that their performances deteriorate as the dimensionality of the search space increases .A variation on the traditional PSO algorithms ,called improved cooperative par‐ticle swarm optimization or COPSO was proposed ,which employed cooperative behavior to significantly improve the performance of the original algorithms .Firstly ,traditional PSO algorithms were used to update the position and velocity vectors of particles , and then a reference global best position was obtained by using cooperative operator on the best positions of particles encountered so far . Such reference global best position was used to update the current global best position of the swarm .Application of the COPSO algorithm on the benchmark optimization problems shows a marked improvement on convergence rate of some traditional PSO algorithms .
出处
《计算机工程与设计》
北大核心
2015年第6期1530-1534,共5页
Computer Engineering and Design
基金
国家档案局2014年科技基金项目(2014-X-65)
达州市2011年科技攻关基金项目(JCY1117)
关键词
PSO算法
协同优化
迭代
全局最优
收敛速度
PSO algorithm
collaborative optimization
iterative
globally optimal
convergence rate
作者简介
侯翔(1983-),男,四川达州人,硕士研究生,讲师,研究方向为计算机网络技术及应用;
蒲国林(1971-),男,四川达州人,博士,副教授,研究方向为服务计算、人工智能。E-mail:dzhouxiang@163.com