Estimation of Platform Waiting Time Distribution Considering Service Reliability Based on Smart Card Data and Performance Reports
Estimation of Platform Waiting Time Distribution Considering Service Reliability Based on Smart Card Data and Performance Reports
复制标题
DOI：
10.3141/2652-04
复制
发表时间：
2017-08
影响因子：
1.7
通讯作者：
A. Wahaballa;F. Kurauchi;Toshiyuki Yamamoto;Jan-Dirk Schmöcker
中科院分区：
文献类型：
--
作者：
A. Wahaballa;F. Kurauchi;Toshiyuki Yamamoto;Jan-Dirk Schmöcker
到目前为止，对站台等候时间的估计很少受到关注。本研究的目的是估计平台上的伦敦地铁等候时间分布，考虑旅行时间的变化，使用智能卡数据，性能报告的补充。假设上车和检票口到站台的步行时间呈正态分布，并与智能卡记录的行程时间相匹配，以估计站台等待时间分布。采用随机前沿面模型，用极大似然法估计模型参数。成本边界函数用于表示智能卡数据中记录的旅行时间作为输出与列车上时间和检票口与站台之间的步行时间作为输入之间的关系。所有估计参数均具有统计学显著性，如p值所示。比较旅行时间的值估计所提出的模型与记录在智能卡数据中的时间显示出良好的拟合系数的决定超过95%。该估计被证明具有快速收敛性和计算效率。研究结果可为公交服务可靠性分析和客流分配提供参考。将获得的分布与观察到的智能卡数据相匹配将有助于估计路线选择行为，从而验证当前的公交分配模型。
The estimation of platform waiting time has so far received little attention. This research aimed to estimate platform waiting time distributions on the London Underground, considering travel time variability by using smart card data that were supplemented by performance reports. The on-train and ticket gate to platform walking times were assumed to be normally distributed and were matched with the trip time recorded by the smart cards to estimate the platform waiting time distribution. The stochastic frontier model was used, and its parameters were estimated by the maximum likelihood method. The cost frontier function was used to represent the relation between the travel time recorded in the smart card data as an output and the on-train time and walking time between the ticket gate and the platform as inputs. All estimated parameters were statistically significant, as shown by p-values. Comparing the travel time values estimated by the proposed model with the times recorded recorded in the smart card data shows a goodness-of-fit coefficient of determination of more than 95%. The estimation proved to have quick convergence and was computationally efficient. The results could facilitate improvements in transit service reliability analysis and passenger flow assignment. Matching the obtained distributions with the observed smart card data will help with estimating route choice behavior that can validate current transit assignment models.