论文部分内容阅读
联盟生成是多Agent系统的一个关键问题,主要研究如何在多Agent系统中动态生成面向任务的最优Agent联盟.引入历史任务集和系统经验集的概念,使用任务相似度来判断任务间的关系.提出了一种基于任务匹配的联盟生成策略,增强了Agent的学习能力,对于任务序列可以有效的求解全局最优联盟.对比实验表明本策略可以有效减少联盟生成的搜索时间和计算量.
Coalition generation is a key issue of multi-agent system, which mainly studies how to dynamically generate task-oriented optimal agent alliance in multi-agent system.It introduces the concepts of historical task set and system experience set, and uses task similarity to judge the relationship between tasks A coalition generation strategy based on task matching is proposed, which enhances the learning ability of the agent and can effectively solve the global optimal coalition for the task sequence.Comparison experiments show that this strategy can effectively reduce the search time and the amount of computation generated by the coalition.