自适应视野和步长的局部邻域人工鱼群算法  

Local neighborhood artificial fish swarm algorithm based on adaptive visual and step

作　　者：许恒迎[1,2] 孙伟斌[1,2] 张霞[1,2] 牛慧娟[1,2] 白成林[1,2] 

机构地区：[1]聊城大学物理科学与信息工程学院,山东聊城252000 [2]山东省光通信科学与技术重点实验室,山东聊城252000

出　　处：《计算机工程与设计》2012年第7期2815-2821,共7页Computer Engineering and Design

基　　金：山东省科技攻关计划基金项目(2009GG10001026)

摘　　要：为解决基本人工鱼群算法搜索后期盲目性大、过早收敛等问题,提出了一种采用全新局部邻域结构的人工鱼群算法。每条人工鱼只能与本邻域内的其他5条邻居鱼通信,每次迭代前每条人工鱼都要根据自身与邻域内其他5条邻居鱼的平均距离自适应地计算视野和步长,并对人工鱼的聚群和追尾行为进行了改进,从理论上讨论了该算法的收敛性。仿真结果和工程实例测试表明,该算法具有良好的收敛速度和全局搜索能力,寻优精度更高,优化性能更好。To solve the large blindness and premature problems of basic artificial fish swarm algorithm in late stage of searching process, an artificial fish swarm algorithm using new local neighborhood structure is proposed, in which each artificial fish can only communicate with other five neighboring artificial fishes. According to the average distance between each artificial fish and other five artificial fishes in the neighborhood, its visual and step are adaptively calculated before each iteration. The swarming and following behaviors of artificial fish is improved. And the convergence of the algorithm is discussed in theory. The simulation results and an engineering example show that the proposed algorithm can converge fast and greatly improve the global searching ability; its optimization precision is even higher and its performance is better.

关 键 词：人工鱼群算法 邻域结构 自适应视野 自适应步长 群体智能 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]