基于驾驶人路径选择偏好的OD行程时间预测方法

以广东省深圳市3 000余辆浮动车近300万组数据为基础, 以地理信息系统技术为主要工具, 以最具代表性的深圳市福田区与罗湖区为研究区域, 确定了不同起讫(OD)点扩展半径。以浮动车唯一编号进行地图匹配, 根据确定的研究区域与扩展半径, 获取了浮动车OD路径与行程时间。确定了驾驶人在进行路径选择时的时间与空间偏好, 建立了基于路径选择偏好的OD行程时间预测方法。以平均绝对百分比误差、均方根相对误差与最大相对误差为指标, 对基于最短路径、最快路径与偏好路径的3种行程时间预测方法进行比较。比较结果表明: 与基于最短路径的预测方法相比, 采用提出方法的平均绝对百分比误差、均方根相对误差与最大相对误差分别降低了66.51%、61.24%、61.47%;与基于最快路径的预测方法相比, 采用提出方法的平均绝对百分比误差、均方根相对误差与最大相对误差分别降低了63.64%、59.70%、58.99%, 因此, 采用基于驾驶人路径选择偏好的OD行程时间预测方法可以显著提高OD行程时间的预测精度。

Prediction method of OD travel time based on driver's route choice preference

Based on about three million data of more than 3 000 floating cars in Shenzhen City of Guangzhou Province, geographic information system (GIS) technology was used as main tool, the representative Futian and Luohu Districts were used as study areas, and the expansion radii of different OD pairs were determined.Map matching was processed by using the unique number of floating car, the OD path and travel time of floating car were obtained according to the determined study area and expansion radius.The driver's temporal and spatial preferences during route choice were determined, and the OD travel time prediction method based on the route choice preference was established.Using the mean absolute percentage error (MAPE), the root mean square relative error (RMSRE), and the maximum relative error (MRE) as indicators, the travel time prediction methods based on the shortest route, the fastest route and the preference route were compared.Comparison result indicates that compared to the prediction method based on the shortest route, the values of MAPE, RMSRE, and MRE of proposed method decrease by 66.51%, 61.24%, and 61.47% respectively, compared to the prediction method based on the fastest route, the values of MAPE, RMSRE, and MRE of proposed method decrease by 63.64%, 59.70%, and 58.99% respectively, so the prediction precision of OD travel time is significantly improved by using the prediction method of OD travel time based on driver's route choice preference.