可求解多目标优化问题的演化博弈优化算法  

Evolutionary Game Based Optimization Algorithm for Multi-objective Optimization Problems

作　　者：徐敏[1] 张敏[1] 王煦法[1] 

机构地区：[1]中国科学技术大学计算机科学与技术系,安徽合肥230027

出　　处：《小型微型计算机系统》2007年第4期640-644,共5页Journal of Chinese Computer Systems

基　　金：国家自然基金委员会海外青年学者合作研究基金(60428202)资助.

摘　　要：借鉴演化博弈的思想和选择机制,提出了一种新的基于演化博弈的优化算法(EGOA)用于多目标问题的求解.算法框架具备对该类问题的通用性.为了对算法性能进行评估,采用了一组多目标优化问题(MOPs)的测试函数进行实验.实验结果表明,使用本算法搜索得到的演化稳定策略集合能够很好地逼近多目标优化问题的帕累托前沿,与一些经典的演化算法相比具有良好的问题求解能力.Used the idea and the selection mechanism called replicator dynamics of evolutionary game theory for reference, and proposed an evolutionary game based optimization algorithm （EGOA） to solve multiobjective optimization problems （MOPs）. The frame of algorithm has universal character for MOPs. To evaluate the newly designed approach, solved a group of test functions. The experiment results showed that the evolutionary stable strategy （ESS） set searched by EGOA is very close to the Pareto front of MOPs. Compared with some classic evolutionary algorithm, has a good ability of problem solving.

关 键 词：多目标优化问题 演化博弈理论 复制者动态 演化算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]