基于双子群和分区采样的果蝇优化新算法  

Novel double subgroups and partition sampling based fruit fly optimization algorithm

作　　者：王友卫[1] 凤丽洲 

机构地区：[1]中央财经大学信息学院,北京100081 [2]天津财经大学理工学院,天津300222

出　　处：《浙江大学学报（工学版）》2017年第11期2292-2298,共7页Journal of Zhejiang University：Engineering Science

基　　金：北京市自然科学基金资助项目(4174105);中央财经大学学科建设基金资助项目(2016XX02)

摘　　要：针对传统果蝇优化算法面临的搜索半径依赖性大、收敛稳定性差、难以协调全局搜索能力及局部搜索能力等问题,提出基于双子群和分区采样的果蝇优化新算法.将果蝇种群划分为搜索果蝇子群和跟随果蝇子群并分别使用这2个子群进行全局化搜索与局部精细化搜索;在每次迭代过程中利用基于分区采样的搜索果蝇位置更新策略,提高算法全局搜索的稳定性;定义了群体聚集度的概念并将其用于协调果蝇的全局搜索能力和局部搜索能力.针对6种典型函数及工业控制系统的测试结果表明,该算法收敛精度高、稳定性好、收敛速度快,与传统算法相比,表现出明显优势.A novel double subgroups and partition sampling based fruit fly optimization algorithm was proposed aiming at the problems that the traditional fruit fly optimization algorithms depend on the searching radius seriously,have bad convergence stability,and cannot balance the global searching ability and local searching ability effectively.The fruit flies were divided into the searching fruit fly subgroup and the following fruit fly subgroup,and these two subgroups were used for global search and local detailed search.A partition sampling based position updating policy of searching fruit fly was used in each iteration process in order to improve the global searching stability of the proposed algorithm.The conception of population aggregation degree was introduced and used to balance the global searching ability and the local searching ability.The test results of six typical test functions and factory control systems show that the proposed method has high searching accuracy,good stability and high convergence speed.The method obtain better performances when compared to traditional algorithms.

关 键 词：果蝇优化算法 搜索果蝇 跟随果蝇 群体聚集度 收敛精度 收敛速度 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]