摘要
粒子群优化算法PSO(Particle Swarm Optimization)目前仍存在着早熟收敛和收敛速度较慢的难题,提出一种新的PSO改进算法。该算法利用水平集对PSO的每一代粒子按照适应度进行划分,把与目标相关的所有信息结合在一起,改变了原有的PSO进化公式,提高了算法的收敛速度;其次,对于每一代的某个个体进行变异,使其变异到粒子密度低的空间中去,从而提高了粒子的多样性,减少早熟发生的机会。实验证明,这种算法是有效的。
Recently there still exist some problems in particle swarm optimization （PSO） algorithm including prematurity and slow convergence. To solve these problems, an improved PSO algorithm based on level set is presented. Particles of each generation are classified, and all the aim-relevant information is arranged effectively. The evolution formula of PSO is changed, and the convergence speed of the algorithm is accelerated. In order to improve the diversity of population and decrease the chance of prematurity, mutation process is carried out for a certain member in each generation and they are moved to room of low-density. Experiment shows that this algorithm is effective.
出处
《计算机应用与软件》
CSCD
北大核心
2008年第4期269-270,275,共3页
Computer Applications and Software
关键词
粒子群优化算法
水平集
密度
Particle swarm optimization（PSO） algorithm Level set Density
