求解多模态函数优化的微果蝇优化算法  

Micro Fly Optimization Algorithm Solving Multi-modal Function Optimization

作　　者：张晓茹[1] 张著洪[2] 

机构地区：[1]贵州大学理学院,贵州贵阳550025 [2]贵州大学大数据与信息工程学院,贵州贵阳550025

出　　处：《信息与控制》2016年第3期361-370,共10页Information and Control

基　　金：国家自然科学基金资助项目(61563009);教育部博士点基金资助项目(20125201110003);贵州大学研究生创新基金资助项目(研理工2015057)

摘　　要：研究求解偏高维多模态函数优化的小种群果蝇优化算法.算法设计中,优质种群经局部变异探测优质个体;中等种群经精英个体引导实现个体转移;劣质种群依赖于精英和劣质个体沿着多方位搜寻多样个体.该算法具有结构简单、可调参数少、进化能力强等优点,其计算复杂度低.比较性的数值实验显示,此算法寻优能力强、搜索效率高且对偏高维函数优化问题具有较好应用潜力.To solve the problem of higher-dimensional multi-modal function optimization,this work investigates a micro-population fly optimization algorithm. In the algorithm design,a local mutation strategy ensures the elitist sub-population to achieve strong exploitation,whereas the elitist individual identified in the process of evolution guides individuals included in the medium sub-population to transform towards specific directions. Moreover,the elitist and worst individuals help the inferior sub-population seek diverse and high-quality individuals along multiple directions. One such algorithm has the merits of structural simplicity,few parameters,strong evolution,and so on. Comparative numerical results show that the algorithm with strong global optimization and high efficiency has great potential for solving higher-dimensional function optimization problems.

关 键 词：果蝇优化 小种群 多模态函数优化 偏高维 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]