多邻域改进粒子群算法  

Multi-neighborhood improved particle swarm optimization algorithm

作　　者：杨雪榕[1] 梁加红[1] 陈凌[1] 尹大伟[2] 

机构地区：[1]国防科学技术大学机电工程与自动化学院,湖南长沙410073 [2]国防科学技术大学航天与材料工程学院,湖南长沙410073

出　　处：《系统工程与电子技术》2010年第11期2453-2458,共6页Systems Engineering and Electronics

基　　金：国家部委级基金(9140A20010409KG0153)资助课题

摘　　要：为了改进标准粒子群算法的性能,提出了多邻域改进粒子群算法。算法提出了一种较为简单的多邻域拓扑方案,对速度惯性权重的更新策略进行了改进,引入了速度和搜索区间限制算法。经过对经典测试函数的计算测试,算法表现出良好的复杂问题求解能力。最后,针对多目标优化问题,给出了多目标应用在粒子群算法中的处理方法,并对经典的5维优化和Golinski减速器设计问题进行了求解,通过数据比对,证明了算法性能远优于现有的一些算法。The multi-neighborhood improved particle swarm optimization algorithm（MNI-PSO） is proposed for the purpose of improving the capability of the standard particle swarm optimization（PSO）.The MNI-PSO contains a simple neighborhood topology and sets an improved update scheme of the velocity inertial weight.The velocity and searching area restriction algorithms are also proposed for the MNI-PSO.The optimization results of the classical testing problems show that the MNI-PSO has performed a great capability for the complex optimization problems.Finally,the solution to multi-objective optimization problems（MOOP） using MNI-PSO is proposed.The classical 5-D optimization problem and Golinski＇s speed reducer problem are optimized.The results show that the MNI-PSO＇s performance is better than some other popular algorithms.

关 键 词：粒子群 多目标优化问题 多邻域拓扑 性能测试 Golinski减速器问题 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]