摘要
本文针对多Agent系统中Agent之间的盲目交互可能产生的效率低下问题。提出了一种基于概念树结构的多Agent合作求解模型。在这个模型中,各Agent基于自己的领域知识构造出概念树,通过Agent之间的合作,对概念树从根节点开始使用证据理论实现逐层聚焦,逐步缩小求解范围。为此,本文基于模态、逻辑和关系概念提出了一种面向可能解集的证据理论表示,并探讨了在多Agent环境下应用证据理论可能导致的若干问题。
Disordered interaction and cooperation among agents could make multi-agent problem solving system inefficient, so a model of multi-agent problem solving based on conceptual tree is proposed. In this model, to solve a problem, every agent would construct a concept tree according to characteristics of problem and corresponding knowledge. According to the hierarchical structure of the concept tree, agents solve the problem, cooperate with each other and synthesize results. Using this method, agents could focus solutions of the problem along the tree. Evidence theory is utilized to synthesize results of deferent agents, and difficulties caused by using evidence theory in MAS are discussed. To overcome these difficulties, an expression of evidence theory based on model logic and conception of relation is proposed.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2003年第3期276-282,共7页
Pattern Recognition and Artificial Intelligence
基金
国家863计划资助项目(No.200lAA110464)
