一种加入局部搜索的双子代差异演化算法  

An improved differential evolution algorithm with local search operator

作　　者：武志峰[1] 刘烁[1] 李耀辉[1] 

机构地区：[1]天津职业技术师范大学信息技术工程学院,天津300222

出　　处：《天津职业技术师范大学学报》2015年第4期1-4,9,共5页Journal of Tianjin University of Technology and Education

基　　金：天津市应用基础与前沿技术研究计划(14JCYBJC15400);天津职业技术师范大学人才计划资助项目(RC14-51)

摘　　要：针对差异演化算法存在收敛速度慢的问题,提出一种加入局部搜索算子的双子代竞争差异演化算法。为了增加群体多样性,对初始种群采用等间隔均匀分布。算法对每一代演化种群中的最优个体以一定概率加入局部随机扰动,在其附近搜索更优秀的新个体,以加快发现最优解的速度。在7个常用测试函数上的实验结果表明:无论是在最优解质量、收敛速度,还是在平均评价次数等方面,本文算法都优于文献[5]的算法。As a particular instance of EA, although differential evolution algorithm is simple and powerful for optimizing continuous functions, it is still faced with premature convergence and easy getting in local optimization problems. In order to solve these problems, an improved differential evolution algorithm with local search operator is presented in this paper. For increasing colony diversity, the population is initialized by an uniform distribution. A local stochastic disturbance term is taken to the best individual of the population in each iteration with a probability. It can search a new individual with better fitness nearby the current individual and speed up the convergence. The experiments on seven common test functions show that this proposed algorithm is superior to the other algorithm on the quality of solution, average evaluation number and con- vergence speed.

关 键 词：差异演化 局部搜索 全局优化 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]