轨道交通站点聚类及其对客流预测的影响分析

doi:10.16097/j.cnki.1009-6744.2023.06.023

交通运输系统工程与信息 ›› 2023, Vol. 23 ›› Issue (6): 227-238.DOI: 10.16097/j.cnki.1009-6744.2023.06.023

轨道交通站点聚类及其对客流预测的影响分析

户佐安^{*1a, 1b, 1c}，邓锦程^1a，杨江浩^1a，赵妍²

1. 西南交通大学，a. 交通运输与物流学院，b. 综合交通大数据应用技术国家工程实验室，c. 综合交通运输智能化国家地方联合工程实验室，成都 611756；2. 成都信息工程大学，资源环境学院，成都 610225

收稿日期:2023-08-02 修回日期:2023-09-27 接受日期:2023-10-07 出版日期:2023-12-25 发布日期:2023-12-23
作者简介:户佐安(1979- )，男，湖北黄梅人，副教授，博士。
基金资助:
国家自然科学基金(61104175)；四川省科技计划项目(2021YJ0067)

Rail Transit Station Clustering and Its Impact on Passenger Flow Forecasting

HU Zuo-an^{*1a, 1b, 1c},DENG Jin-cheng^1a,YANG Jiang-hao^1a,ZHAO Yan²

1a. School of Transportation and Logistics, 1b. National Engineering Laboratory of Integrated Transportation Big Data Application Technology, 1c. National United Engineering Laboratory of Integrated and Intelligent Transportation, Southwest Jiaotong University, Chengdu 611756, China; 2. College of Resources and Environment, Chengdu University of Information Technology, Chengdu 610225, China

Received:2023-08-02 Revised:2023-09-27 Accepted:2023-10-07 Online:2023-12-25 Published:2023-12-23
Supported by:
National Natural Science Foundation of China (61104175)； Sichuan Science and Technology Program (2021YJ0067)

摘要/Abstract

摘要： 城市轨道交通站点受多层面因素交互作用而反映出异质性，为实现站点精细化分类，本文统筹考虑地铁刷卡数据、兴趣点数据和地铁网络数据，提取客流、土地利用和网络性质等特征，其中，客流层面考虑工作日、周末和节假日等不同日期类型下客流状态，土地利用层面考虑站点辐射区用地强度和均衡性，网络层面考虑节点自身特性和影响能力。构建基于主成分分析与K-means++算法的聚类模型，综合聚类评价指标确定簇数，辨析不同类型站点多维度特性，结合站区土地利用和站点网络特征探讨对出行活动的影响，并设计簇内联合预测和整体联合预测策略，采用3种多元时序预测方法探究站点聚类对预测性能的影响。研究结果表明：考虑全部客流特征时，划分为10簇，考虑工作日进站客流特征时，划分为5簇，充分挖掘客流时变特征能够获得更加精细化的聚类结果；各簇站点客流分布特征与其土地利用及网络特征间存在一定的反馈关系；相比于整体联合预测，通过聚类联合相关性强的站点进行预测，以间接捕获空间相关性的方式能有效提升预测性能，各模型均方根误差平均降低9.04%，平均绝对误差平均降低4.94%。研究结果为站点精细化管理和站区设施建设规划提供依据。

关键词: 城市交通, 站点聚类, 机器学习, 轨道交通站点, 多源数据, 客流预测

Abstract: Urban rail transit stations have heterogeneity due to the interaction of multi-level factors. This paper classifies the rail transit stations using the passenger card data, interest point data and subway network data, and extracts passenger flow, land use and network properties. The passenger flow property considers the different status under weekdays, weekends and holidays. The land use property considers the land use intensity and balance of the radiation area of the station. The network property considers the characteristics of nodes and influence capabilities. The paper develops the clustering model of the K-means++ algorithm and uses comprehensive clustering evaluation indicators to determine the number of clusters, distinguish and analyze the multi- dimensional characteristics of different types of stations. It combines the land use of the station area and the characteristics of the station network to analyze the impact on travel activities, and designs the joint prediction within the cluster and the overall joint. Three multivariate time series forecasting methods are used to evaluate the impact of site clustering on forecasting performance. The results show that the rail transit station can be divided into 10 clusters when consider the characteristics of all passenger flow. The rail transit stations can be divided into 5 clusters when consider the characteristics of inbound passenger flow on weekdays. Fully mining the time-varying characteristics of passenger flow can obtain more refined clustering results. There is a certain feedback relationship between passenger flow distribution characteristics and its land use and network characteristics. Compared to the overall joint prediction, the proposed method captures spatial correlation through clustering and joint prediction of sites with strong correlation and can effectively improve the prediction performance. The average reduction in root mean square error was 9.04%, and the average reduction in mean absolute error was 4.94%. The study results provide a basis for the refined management of the station and the planning of station facility construction.

Key words: urban traffic, station clustering, machine learning, rail transit station, multi- source data, passenger flow forecast

中图分类号:

U121

户佐安, 邓锦程, 杨江浩, 赵妍. 轨道交通站点聚类及其对客流预测的影响分析[J]. 交通运输系统工程与信息, 2023, 23(6): 227-238.

HU Zuo-an, DENG Jin-cheng, YANG Jiang-hao, ZHAO Yan. Rail Transit Station Clustering and Its Impact on Passenger Flow Forecasting[J]. Journal of Transportation Systems Engineering and Information Technology, 2023, 23(6): 227-238.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: http://www.tseit.org.cn/CN/10.16097/j.cnki.1009-6744.2023.06.023

http://www.tseit.org.cn/CN/Y2023/V23/I6/227

参考文献

[1] 中国城市轨道交通协会. 城市轨道交通2022年度统计和分析报告[R/OL]. (2023-03-31). [2023-07-20]. https: //www.camet.org.cn/tjxx/11944. [China Urban Rail Transit Association. Urban rail transit 2022 annual statistics and analysis report[R/OL]. (2023-03-31) [2023-07-20]. https://www.camet.org.cn/tjxx/11944.]

[2] 尹芹, 孟斌, 张丽英. 基于客流特征的北京地铁站点类型识别[J]. 地理科学进展, 2016, 35(1): 126-134. [YIN Q, MENG B, ZHANG L Y. Identification of Beijing subway station location based on passenger flow characteristics[J]. Progress in Geography, 2016, 35(1): 126-134.]

[3] 李子浩, 田向亮, 黎忠文, 等. 基于客流规律的地铁车站客流风险分析[J]. 清华大学学报(自然科学版), 2019, 59(10): 854-860. [LI Z H, TIAN X L, LI Z W, et al. Risk analysis of passenger flow in metro stations based on passenger flow law[J]. Journal of Tsinghua University (Natural Science Edition), 2019, 59(10): 854- 860.]

[4] ZHANG L, PEI T, MENG B, et al. Two-phase multivariate time series clustering to classify urban rail transit stations[J]. IEEE Access, 2020, 8: 167998- 168007.

[5] 高勃, 秦勇, 肖雪梅, 等. 基于K-means的北京地铁路网重要度聚类分析[J]. 交通运输系统工程与信息, 2014, 14(3): 207-213. [GAO B, QIN Y, XIAO X M, et al. Clustering analysis of importance of Beijing metro railway network based on K-means[J]. Journal of Transportation Systems Engineering and Information Technology, 2014, 14(3): 207-213.]

[6] XU D, ZHANG X, ZHANG X, et al. Type identification of land use in metro station area based on spatial-temporal features extraction of human activities[J]. Sustainability, 2022, 14(20): 13122.

[7] ZHOU Y, ZHENG S, HU Z, et al. Metro station risk classification based on smart card data: A case study in Beijing[J]. Physica A: Statistical Mechanics and its Applications, 2022, 594(3): 127019.

[8] 蒋阳升, 俞高赏, 胡路, 等. 基于聚类站点客流公共特征的轨道交通车站精细分类[J]. 交通运输系统工程与信息, 2022, 22(4): 106-112. [JIANG Y S, YU G S, HU L, et al. Fine classification of rail transit stations based on the public characteristics of passenger flow at clustered stations[J]. Journal of Transportation Systems Engineering and Information, 2022, 22(4): 106-112.]

[9] LI W, ZHOU M, DONG H. Classifications of stations in urban rail transit based on the two-step cluster[J]. Intelligent Automation and Soft Computing, 2020, 26(3): 531-538.

[10] 李清嘉, 彭建东, 杨红. 武汉市不同站域建成环境与轨道交通站点客流特征关系分析[J]. 地球信息科学学报, 2021, 23(7): 1246-1258. [LI Q J, PENG J D, YANG H. Research on relationship analysis between passenger flow characteristics of rail transit stations and built environment of different station areas in Wuhan[J]. Journal of Geo-information Science, 2021, 23(7): 1246- 1258.]

[11] ZHANG Y, LI Y, ZHANG G. Short-term wind power forecasting approach based on Seq2Seq model using NWP data[J]. Energy, 2020, 213: 118371.

[12] SHAHZADEH A, KHOSRAVI A, NAHAVANDI S. Improving load forecast accuracy by clustering consumers using smart meter data[C]. Killarney: 2015 International Joint Conference on Neural Networks (IJCNN), IEEE, 2015.

[13] ZHANG J, CHEN F, SHEN Q. Cluster-based LSTM network for short-term passenger flow forecasting in urban rail transit[J]. IEEE Access, 2019, 7: 147653- 147671.

[14] 周雨霏, 杨家文, 周江评, 等. 基于热力图数据的轨道交通站点服务区活力测度研究: 以深圳市地铁为例 [J]. 北京大学学报(自然科学版), 2020, 56(5): 875-883. [ZHOU Y F, YANG J W, ZHOU J P, et al. Study on vitality measurement of service area of rail transit station based on heat map data: A case study of Shenzhen metro [J]. Journal of Peking University (Natural Science Edition), 2020, 56(5): 875-883.]

[15] ARTHUR D, VASSILVITSKII S. K-means++ the advantages of careful seeding[C]. New Orleans Louisiana: Proceedings of the Eighteenth Annual ACMSIAM Symposium on Discrete Algorithms, 2007: 1027- 1035.

[16] BAI S, KOLTER J Z, KOLTUN V. An empirical evaluation of generic convolutional and recurrent networks for sequence modeling[J]. ArXiv Preprint, 2018, DOI: arXiv: 1803.01271.

轨道交通站点聚类及其对客流预测的影响分析

Rail Transit Station Clustering and Its Impact on Passenger Flow Forecasting

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	周备, 张莹, 张生瑞, 周千喜, 汪琴. 基于生成对抗网络的追尾事故数据填补方法研究[J]. 交通运输系统工程与信息, 2024, 24(1): 132-137.
[2]	王维莉, 肖雨晴, 周辉, 张为四. 无信控交叉口电动自行车与机动车的冲突博弈研究[J]. 交通运输系统工程与信息, 2024, 24(1): 149-158.
[3]	袁见, 刘福强, 安琨, 郑喆, 马万经, 俞秋田. 基于交通波特征的车道级车流溯源方法[J]. 交通运输系统工程与信息, 2024, 24(1): 159-167.
[4]	胡立伟, 张瑞杰, 赵雪亭, 贺雨, 陈琛, 刘冰, 侯智. 道路环形交叉口机动车冲突风险区识别模型研究[J]. 交通运输系统工程与信息, 2024, 24(1): 168-178.
[5]	张英贵, 赵明慧, 张云丽. 面向虚拟连挂的城轨列车群组追踪运行仿真研究[J]. 交通运输系统工程与信息, 2024, 24(1): 199-209.
[6]	郝妍熙, 卫书彤, 胡华, 王润祺, 刘志钢. 地铁车站步行障碍行人避让行为建模与进出站仿真[J]. 交通运输系统工程与信息, 2024, 24(1): 210-220.
[7]	翁剑成, 张云飞, 林鹏飞, 李文杰. 基于就医活动推断的老年人公交就医可达性研究[J]. 交通运输系统工程与信息, 2024, 24(1): 221-229.
[8]	孙小慧, 黄诚允. 外卖配送电动自行车换电柜选址定容实证研究[J]. 交通运输系统工程与信息, 2024, 24(1): 230-239.
[9]	葛显龙, 王博, 杨育树, 杨昙月, 尹作发. 考虑出行特征的电动汽车协同充电调度优化研究[J]. 交通运输系统工程与信息, 2024, 24(1): 240-252.
[10]	吴静娴, 唐桂孔, 李文翔. 建成环境对共享单车分时骑行量的非线性作用研究：以上海市为例[J]. 交通运输系统工程与信息, 2024, 24(1): 290-298.
[11]	邓毛颖, 邓策方. 基于结构、效率、公平与韧性的路网密度适宜性评价研究[J]. 交通运输系统工程与信息, 2023, 23(6): 22-32.
[12]	荆彬彬, 林永杰, 阎珠豪, 黄政杰. 非闭合交通路网路径选择与绿波协调同步优化方法[J]. 交通运输系统工程与信息, 2023, 23(6): 51-62.
[13]	张兵, 陶文康, 刘建荣, 薛运强, 邓明君. 有限理性视域下老年人出行方式选择巢式Logit模型[J]. 交通运输系统工程与信息, 2023, 23(6): 74-82.
[14]	彭金栓, 袁浩, 宋臻, 范子斌, 杨祥浩, 张磊. 等待阶段使用手机对行人过街风险的影响[J]. 交通运输系统工程与信息, 2023, 23(6): 83-89.
[15]	张文会, 徐海彬, 周舸, 温文. 信控交叉口行人过街冲突严重程度评价模型[J]. 交通运输系统工程与信息, 2023, 23(6): 89-99.