摘要
针对标准粒子群算法在解决多峰问题时极易早熟、局部寻优能力差等缺点,为了达到优化邻域的拓扑结构及增加种群多样性的目的,提出邻居适应值粒子群算法(NFPSO)。该算法将种群分成三个子群,提取每个子群中最优个体,采用加权法构建具有全局代表性的最优个体。通过标准检测函数测试,实验结果表明相比基本粒子群算法,NFPSO具有较好的性能,是对基本粒子群算法的一种有效改进。
In order to overcome the disadvantage of standard particle swarm optimization which has premature convergence when solv- ing the multimodal problem, this paper presents neighbor fitness particle swarm optimization to increase population diversity （NFPSO）. In NFPSO, the swarm is divided into three subgroups where the best individual of each subgroup is selected, and the weighted method is used to structure new gbest. In benchmark functions, the experimental results show that NFPSO has better performance than the basic particle swarm optimization, and it can be regarded as an effective improvement of PSO.
出处
《合肥师范学院学报》
2017年第6期97-100,共4页
Journal of Zunyi Normal University
基金
国家自然科学基金项目(71461027
71471158)
贵州省自然科学基金项目(黔教合KY[2014]295)
关键词
粒子群算法
邻域
全局最优值
particle swarm optimization
neighborhood
global optimum
