基于乘降客流特征的轨道交通站点分类及客流量影响因素分析

Metro Station Classification Based on Boarding and Alighting Passenger Flows and Ridership Impact Factors

既有研究对城市轨道交通客流特征及其影响因素的分析相对成熟,但鲜有对不同类型站点客流量影响因素的研究。本文引入乘降客流特征时间序列聚类法对天津城市轨道交通站点进行分类,基于多源地理大数据从建成环境、社会经济、站点属性与复杂网络特征等维度构建影响因素指标体系,并采用普通最小二乘回归法(OLS)、地理加权回归(GWR)及多尺度地理加权回归(MGWR)这3种回归模型探究不同类型站点客流量影响因素及其影响程度。针对天津的案例,研究表明:基于乘降客流时变特征分类形成居住主导型、就业主导型与商住均衡型这3类站点,各类站点的空间分布及周边土地利用特征均表现出显著差异;对于居住主导型站点客流量影响因素的分析,MGWR模型拟合结果更优,而对于就业主导型与商住均衡型站点客流量影响因素的分析,OLS模型拟合结果较优,但与其他模型相差不大;不同类型站点客流量的显著性影响因素存在差异,不同影响因素对站点客流量的作用方向及强度也存在差异;公交站点密度和开通时长对居住主导型站点客流的影响程度呈现显著的空间异质性特征。研究结果为天津市进一步提升轨道交通运营效能,实现轨道交通站点为主导的综合开发(TOD)提供了分类分区的规划引导策略。

Existing studies have deep analysis on the metro ridership characteristics and its the impact factors,however,the impact factors of passenger flow for different types of metro stations can be further investigated.This paper utilized a time series clustering method which introduced boarding and alighting passenger flow characteristics to classify metro stations in Tianjin city of China,and developed an index system of effect factors according to the built environment,socio-economic,station attributes,and complex network characteristics based on multi-source geographic big data.Three regression models,Ordinary least squares,Geographically Weighted Regression and Multi-Scale Geographically Weighted Regression were used to analyze the factors that affect the ridership and the degree of impactfor different types of stations.The case study in Tianjin indicated that:(1)there are three main categories of stations based on the time-varying characteristics of passenger flow,residential-oriented,employment-oriented and commercial-residential balance stations.The spatial distribution and surrounding land use characteristics of each station were significantly different.(2)For the ridership impact factors at the residential-oriented stations,the MGWR model showed best fitting results.However,for employment-oriented and commercial-residential balance stations,the OLS model demonstrated better fitting results but the differences were marginal compared to other models.(3)The ridership impact factors for different types of stations were significantly different,and the differences were also shown in the direction and intensity of the impact factors.(4)The influence of bus station density and opening hours on the ridership of residential-oriented stations had significant spatial heterogeneity.The study results provided a planning guidance onstation classification and zoning and further improvement of the effectiveness of rail transit operation and development of TOD at rail transit stations in Tianjin.