摘要： 针对动态开放式多智能体系统中联盟的形成问题，提出“按能力分配”的联盟报酬划分规则和联盟报酬率等概念及相关命题，讨论Agent之间的协商机制和协商原则，在此基础上建立基于多智能体遗传算法的联盟形成机制。仿真计算结果表明，该联盟形成机制能够减少系统的通信量，保证所得联盟的稳定性和全局最优性，提高系统的结盟效率。

关键词: 智能体, 多智能体系统, 多智能体联盟, 遗传算法

Abstract: In allusion to the problem of coalition formation in dynamic open Multi-Agent System(MAS), a rule of coalition payoff distribution based on Agents’ abilities is proposed as well as the conception of coalition payoff ratio and other correlative propositions. The negotiation mechanism and negotiation rules among Agents are discussed. Based on these researches, a mechanism to form coalition based on Multi-Agent Genetic Algorithm(MAGA) is established. Simulation results show that this mechanism can reduce the quantity of system’s communications and guarantee the stability and global optimization of coalitions, which can improve the alignment efficiency of system.

Key words: Agent, Multi-Agent System(MAS), multi-Agent coalition, Genetic Algorithm(GA)

中图分类号: 

• TP18