摘要
针对标准粒子群的早熟和局部粒子群的最优位置信息利用率低的问题,提出一类简约的粒子群算法,该算法包含两种改进的策略:初始阶段有区别的更新粒子速度,减少更新频率,当粒子的速度有利于种群的进化时,那么下一代粒子的速度则保持不变;当粒子位置变化不大时,采用基于正态分布的随机采样搜索策略来改变寻优方式,有效地控制种群多样性,避免了早熟现象的发生.仿真实验表明该算法具有更强的寻优能力和更高的稳定性.
In order to avoid premature convergence of PSO （ particle swarm optimization ） and low information utilization of the best positions in the local of PSO, a kind of compact PSO （ CPSO ） algorithm is proposed, in which two strategies are employed： firstly update particle velocity differently in the initial stage, where the velocity can maintain unchanged at the next iteration when it benefits to further improving the fitness. The method not only enhances the local search ability but accelerates particle evolution, then change optimization way with random sampling search strategy based on normal distribution, control the population diversity effectively and avoid premature phenomena. The simulation results show that the algorithm has better probability of finding global optimum and mean best value and can maintain the population diversity in the process of evolution, and it also requires less computation time.
出处
《小型微型计算机系统》
CSCD
北大核心
2012年第4期800-803,共4页
Journal of Chinese Computer Systems
基金
国家"八六三"高技术研究发展计划重大项目(2006AA10A301)资助
江苏省高校自然科学研究项目(10KJD510001)资助
关键词
粒子群算法
早熟
简约
正态分布
particle swarm optimization algorithm
premature
compact
normal distribution
