基于效用理论的有限理性半自治Agent决策方法  

A Decision Making Method in Bounded Rational Semi-autonomous Agent Based on Utility Theory

作　　者：杨克巍[1] 岑凯辉[1] 李孟军[1] 张少丁[1] 

机构地区：[1]国防科学技术大学信息系统与管理学院,长沙410073

出　　处：《华东理工大学学报（自然科学版）》2007年第B06期109-113,共5页Journal of East China University of Science and Technology

摘　　要：为了更加贴近描述agent的决策过程，发挥基于效用理论的决策较之推理决策具有定量化和客观性的优点，通过建立基于半自治作战agent的战场环境，根据有限理性半自治agent的决策过程特点，结合效用理论给出了一种有效的基于效用理论的半自治agent决策方法，将该方法应用于坦克agent对抗中的目标选择模型实例中，通过对仿真结果的分析，表明了此方法适用于建立丰自治agent的决策模型。Decision based on utility theory is more quantitative and objective than reasoning decision. A new method based on utility theory to enhance the decision process in a battlefield combat agent opera-tion simulation is presented. The methodology provides insight into this problem by quantifying force attack, protection, mission failure risk, in the context of the operational threat environment. The target problem in tank engagement simulation is chosen by testing this method, The results show this methods can be applied to build model for semi-autonomous combat agents.

关 键 词：决策 半自治智能体 效用理论 有限理性 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]