一种可用于数值优化的一维智能体遗传算法的研究  

Research of One Dimensional Agent Genetic Algorithm for Global Numerical Optimization

作　　者：李勇明[1] 周頔[2] 曾孝平[1] 

机构地区：[1]重庆大学通信工程学院,重庆400030 [2]四川文理学院,达州635000

出　　处：《系统仿真学报》2009年第1期23-27,共5页Journal of System Simulation

基　　金：国家自然科学基金(30570473);重庆市信息产业局发展资金(200501022);重庆市自然科学基金(CSTC,2008BB2164);重庆大学青年骨干教师创新能力培育基金(CDCX018)

摘　　要：针对复杂寻优问题,提出了一种新的遗传算法-一维智能体遗传算法(CAGA)。该算法采用一维链式智能体网络结构,实现动态邻域竞争选择、正交交叉和自适应变异,可更好的保持种群多样性,从而获得较优的优化精度。实验采用了多个多维复杂函数进行了优化实验,结果表明,该遗传算法比其他多个著名优化算法可获得更优的优化结果。According to complex optimization problem, a new genetic algorithm （one dimensional genetic algorithm） was proposed for global numerical optimization. This algorithm adopts chain-like agent structure, realizes dynamic neighboring competition strategy, orthogonal crossover operation and adaptive mutation operation to keep the diversity of the population better, thereby obtaining better optimization result. Several complex multidimensional benchmark functions were tested for the comparison of CAGA with some other well-known genetic algorithms. The experimental results show that CAGA can obtain better optimization result than some other well-known genetic algorithms.

关 键 词：智能体 遗传算法 数值优化 邻域竞争 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]