交通安全关系千家万户，如何减少交通事故是世界各国都高度关注的问题。Agent具有很多独特的性质，利用Agent可以模拟驾驶员在各种道路环境条件下的行为抉择，挖掘出危险性高的行为，最终为分析研究驾驶员的行为提供参考，这对降低交通事故的发生率无疑是有用的。鉴于此，本文提出一个基于MAS的开放式驾驶员行为决策模型（DBDM）。DBDM以多个决策推理模型的混合推理结果作为驾驶员的策略，并通过对这些策略的不断学习来实现对驾驶员的行为决策过程的模拟。最后，本文还对DBDM进行了计算机仿真，验证了模型的有效性。本文的研究成果还可以为驾驶员处理突发情况提供行为决策支持。

Traffic safety concerns millions of households and the problem about how to reduce traffic accidents has gotten high attention from countries all around the world. Due to the unique features of agent, it is undoubtedly useful to reduce traffic accidents, because it can simulate the driver under the conditions of various road environments and analyze the highrisk behaviors in order to provide reference for the research about drivers’ behaviors. In view of this, the paper presents an open MASbased Driver Behavior Decision Model (DBDM). In the simulation process of reasoning,DBDM can combine and learn the results of reasoning models for the formation of the ultimate strategy. Finally, we also develop a simulation system for DBDM to verify the validity of the model. Our research can also provide decision supports for drivers in emergency.