Abstract: An accurate bus travel time automatic calculation algorithm was proposed based on multi-source data, including the cell tower location trajectory, public transit network and buses’ GPS trajectories. First, the cell tower location trajectory was collected by a smart phone. Second, a sliding time window based algorithm was applied to remove the disturbance data in the cell tower location trajectory caused by cell towers switching “ping-pong effect”. Third, the cell tower location trajectory was matched with the public transit network by their spatial correlation, to obtain locations where the passenger gets on and off the bus. Finally, the cell tower location trajectory was sequence compared with buses’ GPS trajectories, to obtain the passenger’s bus travel time. The experiment of bus travels in Nanhai, Foshan shows that, the error of the bus travel time obtained by the algorithm is within 1 minute, which can satisfy the precision requirement of public transit network adjustment and optimization.

Key words: traffic engineering, travel survey, public transit planning, multi-source data, cell tower location, GPS location

摘要： 提出基于基站定位轨迹、公交线网、公交车GPS轨迹的多源数据精确公交出行时间自动计算方法：第一，利用智能手机采集基站定位轨迹；第二，使用基于滑动时间窗的算法，清除基站定位轨迹中因基站切换“乒乓效应”造成的扰动数据；第三，将基站定位轨迹与公交线网进行空间关联匹配，获得乘客上下车地点；最后，将基站定位轨迹与公交车GPS轨迹进行序列对比，获得乘客公交出行时间。在佛山市南海区进行的公交出行实验表明，该方法获得的公交出行时间误差在1 min以内，能满足公交线网调整与优化的精度要求。

关键词: 交通工程, 出行调查, 公交规划, 多源数据, 基站定位, GPS定位