摘要
微粒群算法是一种新颖的优化算法,已成功应用于许多优化问题,但该算法容易陷入局部极值.针对这种缺陷,提出了一种基于优胜劣汰的多粒子群替代算法,该算法先通过多个种群彼此独立地搜索解空间,增强全局搜索能力;各种群每次进化完成后,核心种群中的最差微粒与其他种群的最好微粒互相替代.通过对3种常用测试函数进行测试和比较,结果表明该算法比标准微粒群算法具有更低的平均最好适应值,可快速收敛到全局最优解,优化效率明显提高.
PSO algorithm is relatively a new optimization algorithm,it has been successfully used in many ptimization problems,but the algorithm is vulnerable to local extreme.One Replacement Optimization of Multi-Particle Swarms is proposed.Particle swarms are employed to search in the solution space independently that enhances the global searching ability.After each particle swarms evolutions,the worst particles are replaced by the other swarm＇s best particles.It makes the particle escaped from the premature convergence and improves the stability of the algorithm.Through testing and comparison with the three kinds of commonly used test functions,the results show that the average of the best fitness of the algorithm is lower than the standard PSO algorithm,it can rapidly converge to the global optimal solution,the optimization efficiency is increased significantly.
出处
《陕西科技大学学报（自然科学版）》
2009年第6期112-115,120,共5页
Journal of Shaanxi University of Science & Technology
关键词
PSO
优化
群智能
多粒子群
PSO
optimization
swarm intelligence
multi-particle swarms
