采用小生境技术的混合蛙跳算法  

A shuffled frog leaping algorithm using niche technology

作　　者：姜建国[1] 李锦[1] 龙秀萍[1] 申洁琳[1] 田旻[1] 

机构地区：[1]西安电子科技大学计算机学院,西安710071

出　　处：《计算力学学报》2012年第6期960-965,共6页Chinese Journal of Computational Mechanics

基　　金：国防基础科研计划(D1120060967)资助项目

摘　　要：混合蛙跳算法是一种新兴的启发式全局优化算法。本文研究了其寻优机制,提出了一种采用小生境技术的混合蛙跳算法:运用RCS小生境技术,使各子种群动态形成了互相独立的搜索空间;在解的更新公式中,设计了一种自适应因子来调节移动步长;采用种群淘汰机制,随机初始化已陷入局部最优的子种群。实验结果表明:本文给出的算法有效提高了寻优精度和收敛速度。Shuffled frog leaping algorithm （SFLA） is a new heuristic global optimization algorithm. After studying the optimization mechanism of SFLA, this paper proposes a shuffled frog leaping algorithm using niche technology. The new algorithm applies the RCS niche technology to make each sub-popula- tion dynamically form search spaces independent of one another. In the update formula for the solution, an adaptive factor is designed to adjust the moving step. Furthermore, the population elimination mecha- nism is used in the algorithm, to randomly initialize the sub-population which falls into the local opti- mum. Experimental results show that the algorithm in this paper improves the optimization accuracy and convergence speed effectively.

关 键 词：混合蛙跳算法 小生境技术 全局优化 种群淘汰 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]