Abstract:

Conventional passenger flow distribution forecasting models for urban rail transit rarely consider the impact of station surrounding area' s land-use, which results in models' accuracies decline dramatically when the land- use around stations changes, so it is necessary to establish the passenger flow distribution forecasting model considering the land-use and their matching degree. First, based on clustering analysis, the matching degree of land-use (MDLU) is defined as an indicator to reflect the correlation between the stations' land-use and passenger flow distribution. Second, the urban rail passenger flow distribution forecasting model is established based on the disaggregate theory, in which, the effects of station surrounding area' s land- use, MDLU, attraction of destination, travel time and etc. on destination choice behavior are considered. Finally, the passenger flow data collected from Guangzhou metro system is used for the case study, and the result shows that the mean absolute errors of the proposed model are successfully limited to 29.30 and 29.52 respectively when the land-use has no-change or change, which demonstrates that the forecasting accuracy of proposed model is satisfactory.

Key words: urban traffic, passenger flow distribution forecast model, disaggregate theory, cluster analysis, land use

摘要：

传统城轨站间客流分布模型较少考虑起讫点土地利用对客流分布的影响，当 站点周边土地利用发生改变时适用性变差，无法保持较好的预测精度，因此有必要构建 考虑站点土地利用及其站点间土地利用性质匹配度的客流分布预测模型.首先，基于站点 聚类分析土地利用性质与客流分布之间的关联性，构造性质匹配度指标；其次，综合考虑 站点土地利用、性质匹配度、终点站吸引量、运行时间等因素建立效用函数，结合客流数 据进行参数估计，构建基于目的地选择的轨道交通客流分布模型；最后，利用广州市轨道 交通客流量数据对其进行精度分析.结果显示，在站点土地利用性质未发生改变与改变情 景下全网站间客流分布量的平均绝对误差仅为29.30 和29.52 人，表明模型具有较高的预 测精度.

关键词: 城市交通, 客流分布模型, 非集计理论, 聚类分析, 土地利用