摘要
分析了实数编码遗传算法处理高维优化问题收敛速度太慢的原因,提出了一种新的种群划分方法模拟生物系统多物种同时进化,指出最优种子的获得不但需要一个好的个体(父体),而且需要一个好的进化方向(母体),通过增加母体的方法加速最优物种的进化;高维数值实验结果验证了该算法的有效性。
On the base of analyzing the reasons for the slow convergence velocity and low convergence of realcode genetic algorithms for high- dimensional optimization problems, the optimal biosystem of current generation was isolated according to Euclid distance between the optimal individual and other individuals and fixing the size of optimal population. The method reducing the bounds of optimal population was adopted. We applied genetic algorithm to two subpopulation with different crossover probability and mutation probability. Obtaining the optimal point not only need a good point, but also need a good evolutionary direction. GA with sex character was used to improve convergence speed of the optimal biosystem. The new approach is compared against other GA in several benchmark functions with high - dimensional optimization problems. The results obtained show that the new approach is a general, effective and robust method.
出处
《武汉理工大学学报》
EI
CAS
CSCD
北大核心
2008年第12期110-113,128,共5页
Journal of Wuhan University of Technology
基金
国家重点973项目(2002CB312203)
高等学校博士学科点专项科研基金项目(20070533131)
关键词
遗传算法
种群划分
物种
性别特征
genetic algorithms
population decomposition
species
sex character
