摘要
粒子群优化算法(PSO)的结构相对简单、运行速度很快,但是算法极易陷入局部最优,出现早熟收敛现象。针对标准粒子群算法存在的问题,引入了一种随迭代次数和粒子间距离大小动态改变的惯性权重,通过设置比例系数控制二者对惯性权重的影响力度。在此基础上为了增加种群多样性,又引入"杂交变异"算子,设计了一种基于杂交变异的动态粒子群优化算法(HV-DPSO)。通过对基准函数的数值试验表明,新算法相对于标准粒子群算法不仅能有效地避免早熟收敛,而且具有更好的收敛效果。
Particle swarm optimization（PSO） is a relatively simple structure which runs very quickly, but it is easily fall into local optimum and appears the phenomenon of premature convergence. Aiming at the PSO existing problems, by setting the proportional coefficient control of inertia weight between influence strength, this paper introduced a kind of novel way using the iteration number and particle size of the distance between the dynamic change inertia weight. At the same time, in order to increase the diversity of population, using ＂hybrid variation＂ operator, designed a kind of dynamic particle swarm optimization based on hybrid variable, （HV-DPSO） based on reference function of numerical experiment. The experimental results show that compared with the traditional PSO, the new algorithm not only can effectively avoid premature convergence but also has better convergence effect.
出处
《计算机科学》
CSCD
北大核心
2013年第11A期143-146,共4页
Computer Science
基金
四川省教育厅青年基金(11ZB058)资助
关键词
粒子群优化算法
动态惯性权重
杂交变异
早熟收敛
多样性
Particle swarm optimization,Dynamic inertial weight,Hybrid variation,premature convergence,Diversity
