基于差分进化的鱼群算法及其函数优化应用  

Fish Swarm Algorithm Based on Differential Evolution and Its Function Optimization Application

作　　者：张大斌[1,2] 杨添柔[1,2] 温梅[1,2] 孙莹[1,2] 周茜[1] 

机构地区：[1]华中师范大学信息管理学院,武汉430079 [2]华中师范大学中科预测科学研究中心,武汉430079

出　　处：《计算机工程》2013年第5期18-22,27,共6页Computer Engineering

基　　金：国家自然科学基金资助项目(70971052)

摘　　要：人工鱼群算法存在收敛速度慢、精度差等不足,借鉴全局的鱼群聚群、追尾行为策略,提出一种基于差分策略的鱼群算法。该算法在鱼群中心执行聚群行为和公告板最优记录的基础上,设置公告板停滞阈值和停滞状态记录,对处于停滞阶段的鱼群进行差分进化操作,进而跳出局部极值,克服后期搜索的无目的性。仿真结果表明,与鱼群算法、粒子群算法进行相比,进化后鱼群算法的收敛速度和寻优精度得到明显改善,具有较好的优化效果。i A novel Artificial Fish Swarm Algorithm（AFSA） based on differential evolution is proposed, which aims to accelerate convergence and improve accuracy of AFSA, and refers to the strategy of globel fish swarm cluster and trace action. The fish cluster for the whole fish center and trace with the bulletin board record in the algorithm. Meanwhile, it is set stagnation threshold and stagnation record in the bulletin board so that fish can execute the differential evolution for out of local minima in the stagnation stages and overcome the lack of purpose of the fish search by it. The convergence and accuracy of the algorithm are improved significantly after evolution. Comparing to the results of other AFSAs and Particle Swarm Optimization（PSO）, result shows that the algorithm has better optimization effects.

关 键 词：鱼群算法 差分进化算法 差分策略 停滞阈值 粒子群优化算法 函数优化问题 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]