动态双子群拟梯度蝙蝠算法  

Dynamic double subgroup pseudo-gradient bat algorithm

作　　者：陈伟利[1] 陈国华[1] 

机构地区：[1]湖南人文科技学院数学与计量经济系,湖南娄底417000

出　　处：《微型机与应用》2016年第6期23-25,28,共4页Microcomputer & Its Applications

摘　　要：针对基本蝙蝠算法存在的易陷入局部最优、后期收敛速度慢等问题,提出动态双子群拟梯度蝙蝠算法。该算法利用蝙蝠脉冲发射频率将蝙蝠种群动态地划分为自由搜索种群和局部搜索种群两个子群,在局部搜索子群中利用拟梯度方向指导蝙蝠搜索。为了验证算法的有效性,通过对4个基准函数的实验测试,实验结果表明,该算法相对于基本蝙蝠算法具有较好的全局搜索能力和优化精度。According to the problems that bat algorithm（BA） is easy to fall into local optimum and converge slowly in the later period,this paper presents an dynamic double subgroup pseudo-gradient bat algorithm,in which,the bat population is divided into free search group and local search group depend on the pulse transmitting frequency. In the local search group,the searching is guided by pseudo-gradient direction.In order to prove the algorithm＇s validity,four typical benchmark functions optimization are tested. The experiment results show that the proposed algorithm has a better global searching capability and optimization precision compared with base bat algorithm.

关 键 词：蝙蝠算法 拟梯度 基准函数 最优值 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]