一种基于排挤机制的改进遗传算法  

Improved Genetic Algorithm Based on Crowding-out Mechanism

作　　者：姜艳姝[1] 梅高铭[1] 

机构地区：[1]哈尔滨理工大学自动化学院,哈尔滨150080

出　　处：《黑龙江水专学报》2009年第1期82-84,共3页Journal of Heilongjiang Hydraulic Engineering College

基　　金：黑龙江省教育厅资助(11511062)

摘　　要：为解决遗传算法求解具有多个极值点的函数时容易出现的早熟的弱点,根据排挤机制,提出了一种改进的遗传算法。该算法采用分组选择的方式,首先按适应度大小进行分组,然后根据个体间相似度来淘汰种群中相似个体,以此维持种群的多样性,避免算法未成熟收敛。用典型函数对算法进行测试,结果表明该算法性能优于基本遗传算法。In order to handle local optima problems when using genetic algorithm to solve the functions with multiple extreme point, the paper puts forward an improved genetic algorithm base on Crowding-out mechanism. The algorithm use the method of grouping of selection, firstly according to individual＇s fitness of the values of the size, divide the population into two groups, then delete the similar individual base on similarity between the group＇s individual so as to maintain the population diversity, and refrain the algorithms from premature convergence. Using typical function to test the algorithms, and the result shows the improved genetic algorithm superior to Simple Genetic Algorithm （SGA） in the algorithmic performance.

关 键 词：遗传算法 早熟收敛 种群多样性 排挤机制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]