文章编号： 1672-5328（2018）02-0083-06

李瑞敏1, 2，陈熙怡1, 2，张睿博3
(1.清华大学交通研究所，北京100084；2.清华大学恒隆房地产研究中心，北京100084；3.廊坊市交通警察支 队，河北廊坊065000)

摘要： 机动车OD矩阵是进行城市道路交通网络分析的核心数据。利用根据车牌识别检测数据分析 得到的道路交叉口转向流量以及整体网络中的部分实测机动车OD信息，使用广义最小二乘模型建 立整合部分机动车OD信息的路网全样机动车OD估计模型，模型中的OD历史值及分配矩阵应用了 真实的部分机动车OD信息推导得到。同时为验证检测数据比例的影响等，使用同一城市两个不同 规模的实际道路网络检测数据，结合S-Paramics 仿真平台对模型进行验证。结果显示，不同的检测 比例对OD估计结果有较为明显的影响，而在较高的检测比例情况下使用转向流量和部分机动车 OD信息可以提高路网全样本机动车OD估计的准确性。

关键词： 交通工程；道路交通；广义最小二乘法；OD估计；转向流量

中图分类号： U491

文献标识码：A

OD Estimation Model Based on Automatic Vehicle Identification Data

Li Ruimin1, 2, Chen Xiyi1, 2, Zhang Ruibo3
(1.Institute of Transportation Engineering, Tsinghua University, Beijing 100084, China; 2.Hang Lung Center for Real Estate, Tsinghua University, Beijing 100084,China; 3.Langfang Traffic Management Bureau, Langfang Hebei 065000, China)

Abstract： Vehicle OD matrix is a critical channel to understand urban road traffic network. Using the intersection turning volumes achieving from license plate recognition data and partial vehicle OD matrix of the whole network, this paper develops an OD matrix estimation model based on the General Least Squares (GLS). The historical OD matrix and assignment matrix are inferred using the real partial vehicle OD matrix. For purpose of evaluating the impacts of different proportion of measured data input to model, the paper uses S-Paramics simulation system to test two different roadway networks within a same city. The results shows that the proportion of measured data exerts a significant impact on the accuracy of OD estimation superficially, higher proportion of measured data, turning volumes and partial vehicle OD information can improve the accuracy of OD estimation.

Keywords： traffic engineering; road traffic; general least squares; OD estimation; turning volumes