摘要
为克服粒子群优化(PSO)易早熟的缺点,提出了一种改进的粒子群优化(MPSO)算法.该算法使整个粒子群按照变异率产生变异粒子,变异的粒子不再朝群体最优解方向飞行,而是朝反方向运动.MPSO提高了种群的多样性,扩大了搜索的空间,提高了粒子群算法摆脱局部最优解的能力.仿真实验表明,改进的粒子群优化算法显著提高了PSO算法的全局搜索能力,且其性能也明显优于遗传算法.
To overcome PSO＇s premature convergence, a modified PSO（MPSO）algorithm is proposed in this paper. It divides the swarm according to the mutation rate. The mutated particles do not fly to the global best solution. In stead, they fly to the reverse direction, thereby increasing the diversity and the exploration space. The MPSO algorithm improves the ability to get out of the local optimization. Emulation experiments demonstrate that the MPSO algorithm remarkablely improves the PSO＇s global search ability, and is superior to genetic algorithm.
出处
《河海大学常州分校学报》
2006年第1期10-13,共4页
Journal of Hohai University Changzhou
基金
湖北省自然科学基金资助项目(2004ABA018)
河海大学常州校区创新基金资助项目(2005B002-01)
关键词
粒子群优化算法
早熟
变异
particle swarm optimization algorithm
premature convergence
mutation
