摘要
针对传统粒子群算法对巨量粒子群收敛时间长且易于陷入局部收敛的现象,提出一种分组协同粒子群优化算法,在巨量粒子群搜索过程中对粒子群进行编组分别独立进化,并针对进化过程中新进的粒子群进行再编组参与进化。在搜索过程中各组的粒子动态的变化,有新加入的粒子也有进入临界区间的粒子,等待各组粒子都进入到临界区间后对所有粒子进行整合再搜索,粒子从分组再到整合可以有效地避免粒子陷入局部收敛,此外再搜索的过程中粒子分布比较集中,则会有较快的收敛速度。文中首先对粒子群模型进行了定义,然后具体描述了粒子群搜索过程中的分组算法,最后对算法进行了验证分析,证明了算法的有效性。
As particle swarm optimization( PSO) algorithm is easy to be lost in local optimal problem and too much time is spent in large swarms searching. This paper proposes the group collaborative particle swarm optimization,grouping the swarms before the search,and re-grouping the swarms in the process of searching. During the searching,each particle dynamically changes. After all particles change into the critical range,search them again. After that,it can avoid particle swarm from being lost in local optimal and at the same time good convergence can be gotten as the particle swarm intensively. The paper defines the model of the particle swarm,and descripts the group algorithm for the particle searching,verifies the algorithm and proves its validity at last.
出处
《信息技术》
2016年第8期14-17,21,共5页
Information Technology
基金
国家自然科学基金项目(61202376)
上海市教委科研创新项目(13YZ075)
