摘要
为了改进标准粒子群优化算法的全局搜索性能,提出了一种种群动态变化的多种群粒子群优化算法。当算法搜索停滞时,把种群分裂成两个子种群,通过子种群粒子随机初始化及个体替代机制增强种群多样性,两个子种群并行搜索一定代数后,通过混合子种群来完成不同子种群中粒子的信息交流。收敛性分析表明,本算法能以概率1收敛到全局最优解。实验结果表明,本算法具有较好的全局寻优能力和较快的收敛速度。
In order to improve the standard particle swarm optimization algorithm global search performance,this paper proposed a novel particle swarm optimization algorithm with population dynamics.When the algorithm search stagnation,divided the population into two sub-populations,obtained population diversity by using random initialization particles and alternative mechanisms of sub-populations in the period of two sub-populations parallel searching.After sub-populations parallel searching,exchanged the information of particle in the different sub-population by mixing two sub-population into one population.The convergence of proposed algorithm was analyzed and the results indicated that it could guarantee converge on the global minimum.The functional test shows that proposed algorithm has better global search ability and fast convergence speed.
出处
《计算机应用研究》
CSCD
北大核心
2011年第3期935-937,共3页
Application Research of Computers
关键词
粒子群优化算法
多种群
种群分裂
种群混合
particle swarm optimization(PSO) algorithm
multi-population
population spliting
population mixing