Comparison of passenger walking speed distribution models in mass transit stations
Comparison of passenger walking speed distribution models in mass transit stations
复制标题
DOI：
10.1016/j.trpro.2017.12.081
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
Xiaoyan Xie;F. Leurent
中科院分区：
文献类型：
--
作者：
Xiaoyan Xie;F. Leurent
在过去的十年中，许多论文集中研究了城市多式联运网络中乘客出行时间的总体变化。遗憾的是，在这大量的研究中，只有少数论文在分解层面上考虑了乘客潜在的步行因素，即步行速度和步行距离。我们最近的研究试图克服这一缺点，通过在沿公共交通线路的一般随机模型中建模均匀分布的步行速度。为了对我们之前的模型进行优化，将一个步行速度为正态分布的新模型M1与之前的步行速度为均匀分布的模型M2和步行速度为固定值的模型M0进行对抗。提出了一种全局比较方法，将这些模型从数值分析、建模和优化框架与实际案例进行比较。对解析公式的数值分析有助于对随机模型的各个部分进行详细的比较。将极大似然估计中M1的一般函数的封闭式公式简化为5段。相反，M2包含17个不同的部分。以巴黎地区最繁忙的快速轨道交通RER A线为例，基于AFC和AVL数据，对其基本分布进行了标准的统计特征分析，得到了较好的模型。该模型将集成到基于AFC和AVL数据的新乘客出行信息模型中。
In the last decade, many papers focused on the study of the variability of passenger journey time in multimodal transport networks in cities on an aggregated level. Unfortunately, among this considerable body of research, only few papers account for passenger underlying walking factors, named walking speed and walking distance, in mass transit stations on disaggregated level. Our recent research tried to overcome this drawback by modelling a uniform-distributed walking speed in a general stochastic model along a mass transit line. To optimize our previous model, a new model M1 with normal-distributed walking speed is confronted with the previous models M2 with a uniform distribution and M0 with fixed value of walking speed. A global comparison approach is proposed to compare those models from numerical analyses, modelling and optimization framework to real case study. Numerical analyses of the analytical formulae hold for detailed comparisons for each part of the stochastic model. The closed-form formula of the general function of M1 in Maximum Likelihood Estimation is reduced to 5 pieces. On the contrary, M2 involves 17 different pieces. The real case study of the busiest express rail transit line RER A in Parisian region is applied based on the AFC and AVL data with standard statistical features analyses of the basic distributions, yielding a better model. This model will be integrated in a new passenger mobility information model based on AFC and AVL data.