基于D-S证据理论的高速公路事件检测信息融合  

Multi-information fusion based on D-S evidence reasoning for freeway incident detection

作　　者：覃频频[1] 许登元[2] 黄大明[3] 

机构地区：[1]西南交通大学交通运输学院,四川成都610031 [2]重庆交通大学计算机与信息学院,重庆400074 [3]广西大学机械工程学院,广西南宁530004

出　　处：《传感器与微系统》2007年第4期24-27,共4页Transducer and Microsystem Technologies

基　　金：国家自然科学基金资助项目(60372065)

摘　　要：为解决高速公路事件检测信息融合问题,在分析D-S证据理论用于多传感器信息融合基本概念的基础上,提出事件检测信息融合D-S方法。具体方法是通过分析高速公路事件检测各种事件信息特点,确定事件检测信息融合对象;再根据事件信息的基本可信度分配,不断融合新的事件信息,更新信任函数,最后,依据基本信任函数的决策方法判断高速公路是否发生事件。来自线圈检测器、移动电话报告和移动车报告3类高速公路事件检测信息融合的实例研究结果表明:通过事件信息融合后的事件检测结果较单一事件信息的结果好,证明事件检测信息融合的正确性和有效性。To resolve freeway incident detection information fusion problems, basic concepts and theory of Dempster-Shafer （D-S） evidence for multi-sensors information fusion is analyzed, and incident detection information fusion D-S method is proposed. Firstly multi-information is compared and incoming fusion information source is determined. Secondly basic probability assignment of incident detection is decided;new information is added continually and believe function is updated. Finally the information fusion result of incident detection is arrived by using the believe function decision method. The information fusion cases from road loop detectors, cellular telephone and probe vehicle show that the information fusion result is better than that of a single information source ;the fusion method is correct and effective in freeway incident detection.

关 键 词：信息处理技术 信息融合 D—S证据理论 事件检测 高速公路 

分 类 号：U491.31[交通运输工程—交通运输规划与管理] TP211[交通运输工程—道路与铁道工程]