多源信息搏弈融合模型的研究及其在ETC中的应用  

Investigation of game-fusion model for multi-source information fusion and its application in ETC

作　　者：廉飞宇[1] 付麦霞[1] 张元[1] 王珂[2] 

机构地区：[1]河南工业大学信息科学与工程学院,河南郑州450001 [2]郑州大学信息工程学院,河南郑州450052

出　　处：《兰州理工大学学报》2007年第3期83-87,共5页Journal of Lanzhou University of Technology

基　　金：国家自然科学基金(60673108);河南杰出人才创新基金(0221000200)

摘　　要：将搏弈论与信息融合技术相结合,针对多源信息冲突环境下的信息融合问题,提出了一种新的通用型融合功能模型多源信息搏弈融合功能模型.该模型在搏弈思想指导下,把局中人、对策和支付作为模型的基本要素,把目标识别或局势评估作为模型的核心功能,把融合中的冲突与合作视为多源信息以不同策略进行的交互搏弈.针对不确定性环境下的态势评估或目标识别,提出了以贝叶斯网络为基础的问题表示及求解方法.在应用中,将博弈融合模型及算法应用于ETC电子不停车收费中的车辆身份识别.实验表明,在车辆身份的识别过程中,博弈融合模型优于传统的信息融合方法,并能有效地改善识别结果,从而验证了博弈融合模型及其算法的有效性和具体应用价值.Aimed at the information fusion in the case of multi-source information conflict and, combining game theory with information fusion together, a multi-source information fusion model based on information game was presented. With the guidance of game theory, the players, strategy, payoff were taken as the basic factors in this model The target identification or situation estimate was taken as its kernel function. And the conflict and cooperation in the fusion were taken as mutual game of multi-source information with different strategies. Aimed at situation estimate or target identification in uncertain conditions, the expressions and solving methods of information game based on Bayesian networks were put forward. In an application, the model and algorithm of information fusion based on information game was applied to vehicle identification in electronic toll collection system. The test result indicated that information fusion based on information game was superior to the traditional information fusion, and could improve identification result remarkably.

关 键 词：信息融合 信息搏弈 贝叶斯网络 电子不停车收费 车辆身份识别 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]