摘要
本文提出了一种带记忆信息的协同进化算法——将种群划分为一个子种群和多个独立的个体,协调算法的局部与全局搜索能力;独立个体中适应度最高的个体与子种群进行交叉与合并,实现种群内部的协作与更新;利用子种群内个体间的相似性,选择有代表性个体进行多次变异,发现有利于提高个体适应度的重要基因位来引导该子种群的变异行为。实验表明,本文算法能够快速找到高精度的数值解,性能稳定且易于实现。
A coevolutionary algorithm with memory （MCEA） is presented to simulate the local self-adaptive evolutionary process of biological colony. Colony is composed of a subgroup and individuals to harmonize preferably global and local search. Crossover and combination are executed between the elite of individuals and the subgroup to coevolve. Comparability among the individuals in the evolution anaphase is utilized. The delegate is executed repeatedly for mutation in order to find the important locas to induct the mutation of subgroups. Experiments show that MCEA is fast, stable and easy-to-realize.
出处
《计算机工程与科学》
CSCD
2008年第3期78-81,共4页
Computer Engineering & Science
关键词
协同进化
自组织学习
记忆型变异
进化算法
coevolution
self-organize study
mutation with memory
evolutionary algorithm
