摘要： 以协同进化遗传算法模拟自动谈判是目前智能计算和多Agent系统等领域研究的新课题。针对现有文献仅模拟单议题谈判的情况，该文提出基于协同进化遗传算法和适应度共享小生境技术的多议题谈判模拟算法，以轮流出价谈判协议为例进行仿真实验。实验结果表明，该算法能在策略种群中形成局部小生境，生成近似Pareto最优的策略集。

关键词: 多议题谈判, 协同进化遗传算法, 小生境, Pareto最优策略

Abstract: Simulating the automated negotiation based on Co-evolutionary Genetic Algorithm(CGA) is a new topic in intelligent computing and multi-Agent system. According to the status that most literatures deal with the single-issue negotiation, this paper puts forward a multi-issue negotiation simulating algorithm based on CGA and fitness sharing Niche technique. Simulation experiment uses the alternating offer protocol as an example to carry out, and validates that the algorithm forms local Niches in strategy populations and produces the set of approximate Pareto optimal strategies set.

Key words: multi-issue negotiation, Co-evolutionary Genetic Algorithm(CGA), Niche, Pareto optimal strategies

中图分类号: 

• TP301.6