基于车牌识别数据的机动车OD估计模型

OD Estimation Model Based on Automatic Vehicle Identification Data

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。

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.