基于车载数据的交叉口车辆行为辨识  

Probe Data-based Identification of Vehicle Behaviors at Intersection

作　　者：姚佼[1] 杨晓光[2] 杨晓芳[1] 安健[3] 

机构地区：[1]上海理工大学管理学院,上海200093 [2]同济大学交通运输工程学院,上海200092 [3]北京交通发展研究中心,北京100073

出　　处：《公路交通科技》2012年第6期127-132,共6页Journal of Highway and Transportation Research and Development

基　　金：国家自然科学基金项目(60974093);同济大学道路与交通工程教育部重点实验室开放基金项目(201107);上海市科委2010年度非政府间国际科技合作项目(10530705100);上海理工大学博士科研启动经费项目(10-10-303-002)

摘　　要：交通控制策略制定过程中,对于车辆在交叉口进口道前行为规律的把握至关重要。准确辨识进口道车辆行为模式,是研究下一代基于车路协同的交通控制理论的基础之一。基于车路实验平台的实时车载数据,运用加权平均法对交叉口的车辆换道和转向行为进行了观察分析,并通过建立线性规划模型及求解对辨识过程中的匹配误差进行了处理。最后实地的案例研究显示,本研究的方法对交叉口驾驶转向行为识别精度达100%,对换道行为识别的误差为8.33%,基本能够满足交通控制的要求。It is important to know vehicle behaviors on control policy. How to identify the vehicle behavior approach of signalized intersection when making traffic mode accurately is one of the fundamental issues to discuss the next generation traffic control theory based on vehicle infrastructure integration. Based on real- time probe data from vehicle-road experimental platform, the lane change and turning behaviors at intersection were observed and analyzed with the weighted average method at first, and then, a linear programming model was established and solved to deal with the matching error in the identification procedure. Finally, with the case study, it is concluded that with vehicle turning movement behavior at intersection identification error is only 8.33% , which is permitted the methods mentioned above, the identification rate of is 100%, in decision and about lane change behavior, the making of traffic control.

关 键 词：交通工程 行为辨识 线性规划 车载数据 车路协同 

分 类 号：U491.23[交通运输工程—交通运输规划与管理]