Prediction of Subway Station Passenger Flow Based on Spatial and Temporal Series Prediction Model

YUAN Shili; XU Haonan; SHEN Luyao; ZHANG Weifeng

doi:10.11907/rjdk.241550

您当前的位置：

首页 >

文章列表页 >

Prediction of Subway Station Passenger Flow Based on Spatial and Temporal Series Prediction Model

Artificial Intelligence | 更新时间：2026-04-30

- Prediction of Subway Station Passenger Flow Based on Spatial and Temporal Series Prediction Model
- Software Guide Vol. 25, Issue 4, Pages: 83-88(2026)
- 作者机构：
  
  南京邮电大学计算机学院，江苏南京 210023
- 作者简介：
- 基金信息：
- DOI：10.11907/rjdk.241550
  CLC： TP183
- Received：18 March 2025，
  
  Published：15 April 2026
- 稿件说明：
移动端阅览
袁世立,徐浩楠,沈路尧等.基于时空序列预测模型的地铁站客流量预测[J].软件导刊,2026,25(04):83-88.

YUAN Shili,XU Haonan,SHEN Luyao,et al.Prediction of Subway Station Passenger Flow Based on Spatial and Temporal Series Prediction Model[J].Software Guide,2026,25(04):83-88.
袁世立,徐浩楠,沈路尧等.基于时空序列预测模型的地铁站客流量预测[J].软件导刊,2026,25(04):83-88. DOI： 10.11907/rjdk.241550.

YUAN Shili,XU Haonan,SHEN Luyao,et al.Prediction of Subway Station Passenger Flow Based on Spatial and Temporal Series Prediction Model[J].Software Guide,2026,25(04):83-88. DOI： 10.11907/rjdk.241550.

摘要

人流量预测在智能交通服务领域发挥着重要作用，交通管理者能够通过提早预测人流量来制定针对性措施，减少拥堵情况发生，缓解交通压力。然而，传统客流量预测通常仅利用历史数据与当前数据的关系进行时间序列预测，精度无法保证。为有效提升预测精度，提出一种时空序列预测模型来预测地铁客流量。该模型采用图卷积神经网络提取各线路之间的空间关系特征，采用长短期记忆网络提取各线路本身的时间特征。利用南京地铁各线路1年来的客流量数据进行验证实验，发现所提模型比传统的ARIMA时间序列预测模型、长短期记忆网络模型的预测准确性更高，尤其在处理非平稳数据时预测误差降低了40%。

Abstract

The prediction of pedestrian flow plays an important role in the field of intelligent transportation services. Traffic managers can develop targeted measures by predicting pedestrian flow in advance， reducing congestion and alleviating traffic pressure. However， traditional passenger flow prediction usually only relies on the relationship between historical data and current data for time series prediction， and the accuracy cannot be guaranteed. To effectively improve prediction accuracy， a spatiotemporal sequence prediction model is proposed to predict subway passenger flow. This model uses graph convolutional neural networks to extract spatial relationship features between each line， and uses long short-term memory networks to extract temporal features of each line itself. Using the passenger flow of various lines of Nanjing Metro over the past year for verification experiments， it was found that the proposed model has higher prediction accuracy than traditional ARIMA time series prediction models and long short-term memory network models， especially when dealing with non-stationary data， with a 40% reduction in prediction error.

关键词

Keywords

references

ZHENG Y C . Full graphic records and subsequent reports of the serious new year stampede on the bund in Shanghai ［N］. Exclusive Net ， 2015-01-01 .

郑渝川 . 上海外滩新年严重踩踏事件图文全记录及后续报道［N］. 独家网， 2015-01-01 .

XIONG Y Y . Urban traffic-flow prediction based on deep learning ［D］. Beijing ： Beijing University of Posts and Telecommunications ， 2020 .

熊奕洋 . 基于深度学习的城市人流量预测［D］. 北京：北京邮电大学， 2020 .

CALABRESE F ， FERRARI L ， BLONDEL D V . Urban sensing using mobile phone network data： a survey of research ［J］. ACM Computing Surveys （CSUR）， 2014 ， 47 （ 2 ）： 1 - 20 . doi: 10.1145/2655691 http://dx.doi.org/10.1145/2655691

VOORT D V M ， DOUGHERTY M ， WATSON S . Combining Kohonen maps with ARIMA time series models to forecast traffic flow ［J］. Transportation Research： Part C ， 1996 ， 4 （ 5 ）： 307 - 318 . doi: 10.1016/s0968-090x(97)82903-8 http://dx.doi.org/10.1016/s0968-090x(97)82903-8

JIAO L ， LI W J ， ZHAO C M ， et al . A season time series （SARIMA） prediction model based on Eviews ［J］. Henan Journal of Preventive Medicine ， 2015 ， 26 （ 5 ）： 349 - 352 . doi: 10.13515/j.cnki.hnjpm.1006-8414.2015.05.008 http://dx.doi.org/10.13515/j.cnki.hnjpm.1006-8414.2015.05.008

焦亮，李文杰，赵超敏，等 . 基于Eviews的季节时间序列（SARIMA）预测模型［J］. 河南预防医学杂志， 2015 ， 26 （ 5 ）： 349 - 352 . doi: 10.13515/j.cnki.hnjpm.1006-8414.2015.05.008 http://dx.doi.org/10.13515/j.cnki.hnjpm.1006-8414.2015.05.008

WANG C C . Short time series based on machine learning ［D］. Changsha ： Hunan University ， 2019 .

汪成成 . 基于机器学习的短时时间序列时空预测方法及应用研究［D］. 长沙：湖南大学， 2019 .

CHANG G ， ZHANG Y ， YAO D Y . Prediction model for short-time traffic flow of regional road network based on spatial otemporal dependence ［J］. Journal of Tsinghua University （Natural Science Edition）， 2013 ， 53 （ 2 ）： 215 - 221 .

常刚，张毅，姚丹亚 . 基于时空依赖性的区域路网短时交通流预测模型［J］. 清华大学学报（自然科学版）， 2013 ， 53 （ 2 ）： 215 - 221 .

WANG S B ， WANG J D ， CHEN H Y . STARMA spatiotemporal sequence prediction model of the network ［J］. Computer Application Research ， 2014 ， 31 （ 8 ）： 2315 - 2319 . doi: 10.3969/j.issn.1001-3695.2014.08.017 http://dx.doi.org/10.3969/j.issn.1001-3695.2014.08.017

王尚北，王建东，陈海燕 . STARMA网络时空序列预测模型研究［J］. 计算机应用研究， 2014 ， 31 （ 8 ）： 2315 - 2319 . doi: 10.3969/j.issn.1001-3695.2014.08.017 http://dx.doi.org/10.3969/j.issn.1001-3695.2014.08.017

SUN S ， ZHANG C ， YU G . A Bayesian network approach to traffic flow forecasting ［J］. IEEE Transactions on Intelligent Transportation Systems ， 2006 ， 7 （ 1 ）： 124 - 132 . doi: 10.1109/tits.2006.869623 http://dx.doi.org/10.1109/tits.2006.869623

CHEN P T ， CHEN F ， QIAN Z . Road traffic congestion monitoring in social media with hinge-loss Markov random fields ［C］// 2014 IEEE International Conference on Data Mining ， 2014 ： 80 - 89 . doi: 10.1109/icdm.2014.139 http://dx.doi.org/10.1109/icdm.2014.139

DAS M ， GHOSH S K . FB-STEP： a fuzzy Bayesian network based data-driven framework for spatio-temporal prediction of climatological time series data ［J］. Expert Systems with Applications ， 2019 ， 117 ： 211 - 227 . doi: 10.1016/j.eswa.2018.08.057 http://dx.doi.org/10.1016/j.eswa.2018.08.057

WANG J Q ， DENG M ， CHENG T . Spatio-temporal sequence data analysis and modeling ［M］. Beijing ： Science Press ， 2012 .

王佳璆，邓敏，程涛 . 时空序列数据分析和建模［M］. 北京：科学出版社， 2012 .

SHI X J ， CHEN Z R ， WANG H ， et al . Convolutional LSTM network：a machine learning approach for precipitation nowcasting ［C］// Proceedings of the 28th International Conference on Neural Information Processing Systems ， 2015 ： 802 - 810 ．

ZHANG J B ， ZHENG Y ， QI D K ， et al . DNN-based prediction model for spatio-temporal data ［C］// Proceedings of the 24th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems ， 2016 ： 92 ． doi: 10.1145/2996913.2997016 http://dx.doi.org/10.1145/2996913.2997016

ZHANG J B ， ZHENG Y ， QI D K . Deep spatio-temporal residual networks for citywide crowd flows prediction［DB/OL］ . https：//arxiv.org/abs/1610.00081 https://arxiv.org/abs/1610.00081 . doi: 10.1609/aaai.v31i1.10735 http://dx.doi.org/10.1609/aaai.v31i1.10735

SHI P Y . Research on prediction methods and their application based on recurrent neural network ［D］. Ji’nan ： Shandong Normal University ， 2020 .

时培莹 . 基于循环神经网络的预测方法及其应用研究［D］. 济南：山东师范大学， 2020 .

LIU H F ， WANG Y R . Research on the short-term power load prediction algorithm based on LSTM ［J］. Modern Information Technology ， 2021 ， 5 （ 23 ）： 40 - 42 .

刘海峰，王艳如 . 基于LSTM的短期电力负荷预测算法研究［J］. 现代信息科技， 2021 ， 5 （ 23 ）： 40 - 42 .

XIE X L . Research and application of cardiopulmonary sound separation method based on non-negative matrix factorization and neural network ［D］. Guangzhou ： Guangdong University of Technology ， 2018 .

谢型浪 . 基于非负矩阵分解和神经网络的心肺音分离方法研究与应用［D］. 广州：广东工业大学， 2018 .

FANG J . Neural network prediction and numerical simulation analysis of subway thermal environment state ［D］. Beijing ： Beijing Jiaotong University ， 2020 .

方杰 . 地铁热环境状态的神经网络预测与数值模拟分析［D］. 北京：北京交通大学， 2020 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Study on Prediction of Soil Greenhouse Gas Emission Based on CNN-LSTM

RF-LSTM Prediction Model for Cyanobacteria Blooms in Dianchi Lake

Research on the Application of Graph Neural Network Model with Update Mechanism in Multiple Object Tracking

Related Author

LI Yarong

YAN Zhengang

GAO Liuyu

ZOU Yang

LIU Yi

DUAN Wei

FAN Siruo

ZHANG Peng

Related Institution

College of Information Science and Technology， Gansu Agricultural University

The Meteorological Bureau of Kunming

Institute of Meteorological Science，Yunnan Meteorological Bureau

College of Computer and Software，Chengdu Neusoft University

College of Intelligent Engineering， Yantai Institute of Science and Technology

⁰