不同类型的城市轨道交通车站客流时空分布特征

Spatial-temporal distribution characteristics of passenger flow in different types of urban rail transit stations

为明晰轨道交通车站功能类型,防范大客流风险和精细化城市管理,探究不同类型车站客流的时空分布特征,采用高斯混合模型(GMM)建立轨道交通车站类型识别方法,运用期望最大化(EM)算法进行求解,选择南京市轨道交通系统进行验证;从进出站时间分布和出行时间分布2个维度,探讨不同类型车站的客流时间分布规律;从车站间的客流起讫点(OD)分布,分析不同类型车站的客流空间分布规律。研究结果表明:南京市128个轨道交通车站可以划分为居住导向型、就业导向型、职住错位型、错位偏居住型、错位偏就业型和枢纽综合型6类;不同类型车站的客流进出站时间分布差异显著,居住导向型和就业导向型车站呈现出典型的单峰形态,进出站客流比介于[0.23,5.59],具有明显的“早进晚出”或“早出晚进”客流高峰;职住错位型车站呈现出典型的双峰形态,进出站客流比分别为1.19和1.07,早晚高峰时段的进出客流较为均衡;错位偏居住型和错位偏就业型车站也呈现出双峰形态,但2个峰值大小不同;枢纽综合型车站没有显著的进出站客流早晚高峰,客流波动没有明显的规律性;不同类型车站的进出站早晚高峰时段不一致,其中早高峰时段出站时间的高峰比进站时间晚15~45 min;不同类型车站乘客出行时间均值和标准差不同,但居住导向型和错位偏居住型车站出发客流的出行时间均值和标准差均高于其他车站,体现出长时间通勤特征;车站空间分布具有明显的圈层式结构,以主城区为圆心,车站从内向外依次为就业导向型、错位偏就业型、职住错位型、错位偏居住型和居住导向型;车站间的OD客流具有明显的分级特征,流量排名前20%的OD线路占到了总客流的79.01%,符合典型的“二八定律”。

In order to clarify the functional types of rail transit stations, avoid large passenger flow risks and refine urban management, the spatial-temporal distribution characteristics of passenger flow in different types of stations was explored. Gaussian mixture model(GMM)was used to establish the identification method of rail transit station type, and expectation maximization(EM)algorithm was utilized to solve the problem. Nanjing rail transit system was selected to verify the effectiveness of the proposed method. The time distribution regularity of passenger flow at different stations was explored from the perspective of time distribution of checking in and out passenger flow and travel time. By using the passenger flow OD distribution between stations, the spatial distribution regularity of passenger flow at different stations was analyzed. The results show that 128 rail transit stations in Nanjing can be divided into six types, namely residential-oriented station, employment-oriented station, spatial mismatched station, mixed mainly residential-oriented station, mixed mainly employment-oriented station and hub comprehensive station. Significant differences can be found in the time distribution of the passenger flow in different stations. Residential-oriented and employment-oriented stations show a typical single peak shape, with a passenger flow ratio between [0.23,5.59] and a clear passenger flow peak of “early in and late out” or “early out and late in”. Spatial mismatched stations present a typical bimodal pattern, with a balanced passenger flow in the morning and evening peak hours, and a passenger flow ratio of 1.19 and 1.07, respectively. Mixed mainly residential-oriented stations and mixed mainly employment-oriented stations also present a bimodal pattern, but the sizes of the two peaks are different. There is neither significant morning and evening checking in and out passenger flow peak nor obvious regularity in passenger flow fluctuations at hub comprehensive stations. The checking in and out peak hours of different types of stations are inconsistent. Compared with the peak of the arrival time, there is a delay of 15 to 45 minutes about departure time during the morning and evening peak hours. The mean and standard deviation of passenger travel time vary among different types of stations. However, the mean and standard deviation of travel time at residential-oriented and mixed mainly residential-oriented stations are higher than those at other stations, reflecting the characteristics of the long-term commuting. The spatial distribution of stations presents an obvious ring structure. With the main city as the center of the circle, the station type from inside to outside in turn are employment-oriented station, mixed mainly employment-oriented station, spatial mismatched station, mixed mainly residential-oriented station, and residential-oriented station. The passenger flow OD between stations are obviously different. The top 20% OD lines account for 79.01% of the total passenger flow thus conforming to the typical Pareto's law.3 tabs, 10 figs, 23 refs.