双链式智能体遗传算法用于数值优化的研究  

Research of Double-ring Agent Genetic Algorithm for Global Numerical Optimization

作　　者：伍文平[1,2] 魏明[1,2] 王东[1,2] 王刚[1,2] 

机构地区：[1]解放军69046部队仿真技术研究室,乌鲁木齐830002 [2]重庆大学通信工程学院,重庆400030

出　　处：《科学技术与工程》2008年第10期2523-2529,共7页Science Technology and Engineering

基　　金：国家自然科学基金(70472011)资助

摘　　要：针对传统智能体遗传算法全局优化计算精度不够高、时间较长的不足,提出了一种改进的双链式智能体结构,并基于此设计了一种新的智能体遗传算法——双链式智能体遗传算法。该算法采用了多子群并行搜索的模式,闭合链式智能体结构和循环链式智能体结构,可实现多机并行优化,具有优化时间短、优化精度高的特点。为了验证本文算法的优越性,采用国际标准的测试函数对该算法性能进行测试,并与智能体遗传算法(MAGA)相比较。实验结果表明,该算法在全局优化精度、优化收敛速度方面均优于MAGA。For the low optimization precision and long optimization time of classical agent genetic algorithm, double chain-like agents structure is proposed to construct a kind of multi-population agent co-genetic algorithm with chain-like agent structure （DCAGA）. This algorithm adopted multi-population parallel searching mode, close chain-like agent structure, cycle chain-like agent structure, and has the characteristics of high optimization precision and short optimization time. For verifying this algorithm, some popular benchmark functions were used for test this algorithm and a kind of popular agent genetic algorithm （MAGA）. The experimental results show that DCAGA has higher optimization precision and shorter optimization time than MAGA.

关 键 词：遗传算法 双链式 智能体 数值优化 

分 类 号：O224[理学—运筹学与控制论]