基于柯西变异的果蝇优化算法  

Fruit Fly Optimization Algorithm Based on Cauchy Mutation

作　　者：郭德龙[1] 杨楠[1] 周永权[2] Guo Delong;Yang Nan;Zhou Yongquan(Department of Maths Qiannan Normal University for Nationalities,Duyun 558000,China;College of Information Science and Engineering,Guangxi University for Nationalities,Nanning 530006,China)

机构地区：[1]黔南民族师范学院数学与统计学院,贵州都匀558000 [2]广西民族大学信息科学与工程学院,广西南宁530006

出　　处：《科技通报》2017年第3期223-228,257,共7页Bulletin of Science and Technology

基　　金：国家自然科学基金项目(61463044);国家自然科学基金项目(61165015);国家民委科研基金项目(0832082);贵州省科技厅联合基金(黔科LH[2014]7436);广西复杂系统与智能计算重点实验室开放课题(15CI04Y)

摘　　要：针对果蝇优化算法是模仿果蝇寻找食物行为而进行全局搜索最优解的新算法,该算法存在容易陷入局部最优解和收敛速度慢的缺点。提出了一种基于柯西变异的果蝇优化算法,利用柯西分布具有较高的两翼概率特性从而容易产生一个远离原点的随机生成数,即柯西分布有一条很长的尾巴。所以在果蝇个体利用嗅觉搜索食物之随机方向距离上引入柯西变异算子代替原来随机方向变异算子进行扰动,从而容易跳出局部最优。最后通过数值仿真实验对6个标准测试函数来进行作对比检验,结果表明该算法在求解高维函数优化问题更好。Based on the shortcomings of Fruit Fly Optimization Algorithm,which imitates the behavior of flies looking for food,such as low precision,slow convergence rate,and easily falling into local optimal solution,an improvement about the Fruit Fly Optimization Algorithm that is based on Cauchy mutation is put forward.Cauchy distribution has a high probability of the two wings characteristics,which can easily generate a randomly generated number away from the origin.That is,Cauchy distribution has a long tail.This method applies Cauchy mutation operator instead of the original direction of random mutation operator into the random directions of the individual fruit fly’s sense of smell while looking for food,thus jumping out of local optimum.At last,a contrast test is conducted to compare the six standard test functions,with the results showing that this algorithm is much better in solving the high-dimensional function optimization problems.

关 键 词：果蝇优化算法 柯西分布 变异算子 收敛 柯西变异 高维函数 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]