摘要： 多Agent系统存在的动态特性使证据推理中的可传递置信模型(TBM)能够有效地处理动态环境的证据推理。在分析和研究可传递置信模型算法的基础上，提出一种基于证据推理TBM模型的多Agent决策融合方法，构建多Agent决策融合系统的框架模型，分析该系统的信息更新、合成算法及决策制定算法。利用SimuroSot作为仿真平台，将该方法应用于判断对手的队形和策略，得到了较满意的结果。

关键词: 证据推理, 可传递置信模型, 多Agent系统, 决策融合

Abstract: The Transferable Belief Model(TBM) of evidential reasoning is able to resolve the dynamic context’s evidential reasoning effectively aiming at the Multi-Agent System(MAS)’s dynamic property. This paper presents the multi-Agent decision fusion system based on TBA, and builds the multi-Agent decision fusion system model. It discusses this system’s algorithm of information refresh, combination and decision-making. With the simulation platform of SimuroSot, this method is applied to the situation assessment of match and provides better results.

Key words: evidential reasoning, Transferable Belief Model(TBM), Multi-Agent System(MAS), decision fusion

中图分类号: 

• TP181