基于手机应用软件测算OD数据的区域交通需求预测

Regional Traffic Demand Forecasting Based on OD Data Calculated by Mobile Phone Application Software

随着QQ、微信、支付宝为代表的手机应用在线上和线下各领域的推广和普及,基于位置的服务数据在获取和分析人口出行信息等方面不断体现出很好的应用前景。为探究手机应用软件测算的OD交通量在预测区域交通需求中的可行性,基于腾讯提供的OD交通量数据,以京津冀城市群为例,对城市群日均交通出行量进行了测算和分析。测算方法以传统的交通分配四阶段法为基础,通过设定交通小区划分标准、选取小区质心点、建立路段阻抗计算模型及流量转换模型,在交通规划软件中实现了京津冀城市群主要公路网交通流量的分配。为验证流量分配结果,研究选取了8条代表线路的流量分配结果与2016年实测的交通调查数据进行了比较。结果表明:观测流量与分配流量的误差率在可接受范围内波动,且两者具有很好的线性相关性,可决系数为0.95,从而验证了利用手机多源位置数据进行交通需求预测的可行性。对此类数据在大区域交通规划、出行需求预测、多方式交通组合优化、运行分析方面提出了进一步的应用建议。

With the promotion and popularization of mobile phone applications in online and offline fields such as QQ, WeChat, Alipay, it shows a good application prospect continuously for location-based service(LBS) data in obtaining and analyzing population trip information. In order to investigate the feasibility of using Origin-Destination(OD) traffic data calculated by mobile phone applications to forecast regional travel demand, by using the OD traffic data provided by Tencent, the daily average traffic volume of the road network of the Beijing-Tianjin-Hebei city cluster is predicted and analyzed. The prediction is based on the traditional traffic assignment method: "four-stage method", which includes setting division criterion of traffic analysis zones, selecting centroids, establishing models of link impedance calculation and traffic conversion. Then the traffic assignment in the key road network of the Beijing-Tianjin-Hebei city cluster is realized by using a traffic planning software. In order to verify the result of the traffic assignment, the traffic assignment of 8 representative roads is selected to be compared with the field traffic survey data in 2016. The result shows that the error rate between the predicted volume and the survey data of traffic volume in 2016 fluctuates within acceptable range, and there is a good linear correlation between these 2 kinds of volume, the determination coefficient is 0.95, which shows great liner correlation, which verified the feasibility of using mobile phone multi-source applications’ location data to forecast traffic demand. The suggestions for the application of these data in transport planning for large area, travel demand forecasting, multi-modal traffic optimization and traffic operation analysis are proposed.