Passenger Flow Prediction for Public Transportation Stations Based on Spatio-Temporal Graph Convolutional Network with Periodic Components

Weixi HUANG, Baokui CAO, Xueming LI, Mingliang ZHOU*

*Corresponding author for this work

Research output: Journal PublicationsJournal Article (refereed)peer-review

5 Citations (Scopus)

Abstract

Station passenger float forecasting is a normal spatio-temporal statistics forecasting problem. Effectively capturing comprehensive spatio-temporal correlations in such data plays a key role in solving such problems. This paper proposes Spatio-Temporal Graph Convolutional Neural Network Based on Periodic Component (Periodic ST-GCN) to predict the passenger glide at public transportation stations. The model now not solely captures the spatio-temporal traits of visitors' facts through the spatio-temporal convolution block with a sandwich shape composed of one spatial-dimensional convolution and two temporal dimensional convolutions. Also, it effectively considers the periodicity of passenger flow at public transportation stations through the recent, daily and weekly periodic components and, because the graph convolution in the spatial dimension uses pure convolution operations, it reduces the model' training parameters and converges faster. Through the experiment of predicting the Origin-Destination (OD) of passenger flow at public transportation stations in Chongqing, it is found that Periodic ST-GCN achieves better results in two evaluation indicators, mean absolute error (MAE) and root mean square deviation (RMSE).

Original languageEnglish
Article number2250134
JournalJournal of Circuits, Systems and Computers
Volume31
Issue number7
Early online date26 Jan 2022
DOIs
Publication statusPublished - 15 May 2022
Externally publishedYes

Bibliographical note

Publisher Copyright:
© 2022 World Scientific Publishing Company.

Keywords

  • Passenger flow prediction
  • periodic components
  • spatio-temporal graph convolutional network

Fingerprint

Dive into the research topics of 'Passenger Flow Prediction for Public Transportation Stations Based on Spatio-Temporal Graph Convolutional Network with Periodic Components'. Together they form a unique fingerprint.

Cite this