- 规划与建设
- 文章编号：1009-6000(2026)06-0078-09
- 中图分类号：U291 文献标识码：B
- Doi：10.3969/j.issn.1009-6000.2026.06.011
- 项目基金：自然资源部城市国土资源监测与仿真重点实验室开放基金资助课题“国土空间规划全生命周期在线管理的关键技术研究”（K F-2023-08-11）；国家自然科学基金重大项目“国土空间多要素协同的智能规划决策技术与验证”（62394335）。
- 作者简介：辜智慧，自然资源部城市国土资源监测与仿真重点实验室，深圳大学建筑与城市规划学院，教授，博士生导师，主要从事城乡空间信息技术与应用研究；
边梦圆，深圳大学建筑与城市规划学院，硕士研究生；
罗成，深圳市福田区城市管理和综合执法局，副科长；
段利鹏，深圳大学建筑与城市规划学院，博士研究生；
张艳，通信作者，深圳大学建筑与城市规划学院，教授，博士生导师，主要从事交通与土地使用一体化研究。
- 轨道站域间通勤行为的轨道分担率及其影响机制研究——基于手机信令数据的深圳市实证分析
- Rail Transit Mode Share in Inter-Station Commuting and Its Influencing Mechanisms: An Empirical Analysis of Shenzhen Based on Mobile Signaling Data
- 辜智慧 边梦圆 罗成 段利鹏 张艳
- GU Zhihui BIAN Mengyuan LUO Cheng DUAN Lipeng ZHANG Yan
- 摘要：
轨道分担率是揭示居民出行行为中轨道交通吸引力的重要指标，然而居民出行的轨道分担率特征及其影响机制仍然没有得到充分研究。研究以深圳市为例，利用手机信令识别居民出行行为并判断其是否包含轨道出行；提取了 141 个地下轨道站点 800 m 范围内早高峰站域间通勤数据并计算其轨道分担率，采用极致梯度提升决策树分析了站域及线路层面不同因素的影响，结果表明：1）低轨道分担率以原关内短距离通勤为主，高轨道分担率以原关内东西向长距离通勤和原关内外中长距离通勤为主；2）轨道分担率在站域联系层面主要的影响因素是站点间距离、总出行人次和非直线系数，在站点属性层面主要的影响因素有 D 站（destination station）就业人口、O 站（origin station）工业建筑面积、O 站居住人口、D 站商业建筑面积、O 站商业建筑面积以及O 站平均房屋租金，上述 9 个因素的重要度达 60%；3）这些影响因素与轨道分担率呈现明显的非线性关系，并存在阈值效应，同时，部分影响因素之间可以相互验证，且对轨道分担率的影响存在交互作用。研究认为提升轨道线路与职住关联的空间适配、合理布局轨道站域的建筑功能配比和增加高需求区段的发车频次是优化居民通勤出行轨道分担率的关键。 - 关键词：
轨道交通；早高峰通勤；轨道分担率；极致梯度提升决策树；手机信令数据 - Abstract: Rail transit mode share is a critical indicator for revealing the attractiveness of rail transit in residents’ travel behavior. However, the characteristics of rail transit mode share in residents’ commuting and its underlying influencing mechanisms remain underexplored. Taking Shenzhen as a case study, this study utilizes mobile signaling data to identify residents’ travel behavior and determine whether rail transit is involved. It extracts morning peak hour commuting trips between station catchment areas within an 800 meter radius of 141 underground rail stations and calculates the corresponding rail transit mode share. The Extreme Gradient Boosting (XGBoost) model is employed to analyze the effects of various factors at both the station-area and line levels. The results show that: (1) Low rail transit mode share is predominantly observed in short-distance commuting within the former Special Economic Zone (SEZ), while high rail transit mode share is mainly found in east-west long-distance commuting within the former SEZ and in medium- to long-distance commuting between the former SEZ and the areas outside it. (2) At the station-pair level, the primary influencing factors of rail transit mode share are inter-station distance, total number of commuting trips, and the nonlinear coefficient; at the station-attribute level, the key factors include employment population at destination stations, industrial floor area at origin stations, residential population at origin stations, commercial floor area at destination stations, commercial floor area at origin stations, and housing rent at origin stations. The cumulative importance of these nine factors amounts to 60%. (3) These influencing factors exhibit significant nonlinear relationships with rail transit mode share, with threshold effects. Moreover, some factors mutually validate each other and exert interactive effects on rail transit mode share. The study concludes that enhancing the spatial alignment between rail transit lines and the distribution of jobs and housing, rationally allocating the mix of building functions within station catchments, and increasing train frequencies in high-demand sections are key to optimizing rail transit mode share for residents’ commuting trips.
- Key words: rail transit; morning peak commuting; rail transit mode rate; XGBoost; mobile signaling data