A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift

Yong Liu; Jianxun Cui; Zhaohua Long

doi:10.1007/978-3-031-70507-6_15

A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift

Yong Liu
, Jianxun Cui^*
, Zhaohua Long

^*Corresponding author for this work

School of Transportation Science and Engineering, Harbin Institute of Technology
Chongqing Research Institute of HIT
Ltd.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Accurate prediction of short-term traffic flow is crucial for the control and guidance of urban traffic. This paper proposes a new deep learning traffic flow prediction model, Spatial-Temporal fusion model based on Deformable Convolution (DConv-ST), which deeply captures the spatiotemporal correlations present in the sequences. The model divides the raw data into three types of sequences containing information about recent, daily, and weekly periods. A deformable convolutional module is constructed to solve the problem of temporal periodic drift in the sequence, and graph attention network and multi-head attention mechanism are used to capture local and global spatial correlations. A gated mechanism is used to fuse the results of each component module for output. Experiments including model performance analysis, important component analysis, and ablation analysis were conducted on the publicly available transportation network datasets PeMS04 and PeMS08. The experimental results all demonstrate the superior performance of the proposed model.

Original language	English
Title of host publication	IoT as a Service - 9th EAI International Conference, IoTaaS 2023, Proceedings
Editors	Xiang Chen, Xijun Wang, Shangjing Lin, Jing Liu
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	183-199
Number of pages	17
ISBN (Print)	9783031705069
DOIs	https://doi.org/10.1007/978-3-031-70507-6_15
State	Published - 2025
Externally published	Yes
Event	9th EAI International Conference on IoT as a Service, IoTaaS 2023 - Nanjing, China Duration: 27 Oct 2023 → 29 Oct 2023

Publication series

Name	Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST
Volume	585 LNICST
ISSN (Print)	1867-8211
ISSN (Electronic)	1867-822X

Conference

Conference	9th EAI International Conference on IoT as a Service, IoTaaS 2023
Country/Territory	China
City	Nanjing
Period	27/10/23 → 29/10/23

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 11 Sustainable Cities and Communities

Keywords

Attention mechanism
Short-term traffic flow prediction
Temporal periodic drift

Access to Document

10.1007/978-3-031-70507-6_15

Cite this

Liu, Y., Cui, J., & Long, Z. (2025). A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift. In X. Chen, X. Wang, S. Lin, & J. Liu (Eds.), IoT as a Service - 9th EAI International Conference, IoTaaS 2023, Proceedings (pp. 183-199). (Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST; Vol. 585 LNICST). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-70507-6_15

Liu, Yong ; Cui, Jianxun ; Long, Zhaohua. / A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift. IoT as a Service - 9th EAI International Conference, IoTaaS 2023, Proceedings. editor / Xiang Chen ; Xijun Wang ; Shangjing Lin ; Jing Liu. Springer Science and Business Media Deutschland GmbH, 2025. pp. 183-199 (Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST).

@inproceedings{6d60cfe0227e41d9af6a1c9f76b82f2c,

title = "A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift",

abstract = "Accurate prediction of short-term traffic flow is crucial for the control and guidance of urban traffic. This paper proposes a new deep learning traffic flow prediction model, Spatial-Temporal fusion model based on Deformable Convolution (DConv-ST), which deeply captures the spatiotemporal correlations present in the sequences. The model divides the raw data into three types of sequences containing information about recent, daily, and weekly periods. A deformable convolutional module is constructed to solve the problem of temporal periodic drift in the sequence, and graph attention network and multi-head attention mechanism are used to capture local and global spatial correlations. A gated mechanism is used to fuse the results of each component module for output. Experiments including model performance analysis, important component analysis, and ablation analysis were conducted on the publicly available transportation network datasets PeMS04 and PeMS08. The experimental results all demonstrate the superior performance of the proposed model.",

keywords = "Attention mechanism, Short-term traffic flow prediction, Temporal periodic drift",

author = "Yong Liu and Jianxun Cui and Zhaohua Long",

note = "Publisher Copyright: {\textcopyright} ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2025.; 9th EAI International Conference on IoT as a Service, IoTaaS 2023 ; Conference date: 27-10-2023 Through 29-10-2023",

year = "2025",

doi = "10.1007/978-3-031-70507-6\_15",

language = "英语",

isbn = "9783031705069",

series = "Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "183--199",

editor = "Xiang Chen and Xijun Wang and Shangjing Lin and Jing Liu",

booktitle = "IoT as a Service - 9th EAI International Conference, IoTaaS 2023, Proceedings",

address = "德国",

}

Liu, Y , Cui, J & Long, Z 2025, A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift. in X Chen, X Wang, S Lin & J Liu (eds), IoT as a Service - 9th EAI International Conference, IoTaaS 2023, Proceedings. Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST, vol. 585 LNICST, Springer Science and Business Media Deutschland GmbH, pp. 183-199, 9th EAI International Conference on IoT as a Service, IoTaaS 2023, Nanjing, China, 27/10/23. https://doi.org/10.1007/978-3-031-70507-6_15

A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift. / Liu, Yong ; Cui, Jianxun; Long, Zhaohua.
IoT as a Service - 9th EAI International Conference, IoTaaS 2023, Proceedings. ed. / Xiang Chen; Xijun Wang; Shangjing Lin; Jing Liu. Springer Science and Business Media Deutschland GmbH, 2025. p. 183-199 (Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST; Vol. 585 LNICST).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift

AU - Liu, Yong

AU - Cui, Jianxun

AU - Long, Zhaohua

N1 - Publisher Copyright: © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2025.

PY - 2025

Y1 - 2025

N2 - Accurate prediction of short-term traffic flow is crucial for the control and guidance of urban traffic. This paper proposes a new deep learning traffic flow prediction model, Spatial-Temporal fusion model based on Deformable Convolution (DConv-ST), which deeply captures the spatiotemporal correlations present in the sequences. The model divides the raw data into three types of sequences containing information about recent, daily, and weekly periods. A deformable convolutional module is constructed to solve the problem of temporal periodic drift in the sequence, and graph attention network and multi-head attention mechanism are used to capture local and global spatial correlations. A gated mechanism is used to fuse the results of each component module for output. Experiments including model performance analysis, important component analysis, and ablation analysis were conducted on the publicly available transportation network datasets PeMS04 and PeMS08. The experimental results all demonstrate the superior performance of the proposed model.

AB - Accurate prediction of short-term traffic flow is crucial for the control and guidance of urban traffic. This paper proposes a new deep learning traffic flow prediction model, Spatial-Temporal fusion model based on Deformable Convolution (DConv-ST), which deeply captures the spatiotemporal correlations present in the sequences. The model divides the raw data into three types of sequences containing information about recent, daily, and weekly periods. A deformable convolutional module is constructed to solve the problem of temporal periodic drift in the sequence, and graph attention network and multi-head attention mechanism are used to capture local and global spatial correlations. A gated mechanism is used to fuse the results of each component module for output. Experiments including model performance analysis, important component analysis, and ablation analysis were conducted on the publicly available transportation network datasets PeMS04 and PeMS08. The experimental results all demonstrate the superior performance of the proposed model.

KW - Attention mechanism

KW - Short-term traffic flow prediction

KW - Temporal periodic drift

UR - https://www.scopus.com/pages/publications/85209578059

U2 - 10.1007/978-3-031-70507-6_15

DO - 10.1007/978-3-031-70507-6_15

M3 - 会议稿件

AN - SCOPUS:85209578059

SN - 9783031705069

T3 - Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST

SP - 183

EP - 199

BT - IoT as a Service - 9th EAI International Conference, IoTaaS 2023, Proceedings

A2 - Chen, Xiang

A2 - Wang, Xijun

A2 - Lin, Shangjing

A2 - Liu, Jing

PB - Springer Science and Business Media Deutschland GmbH

T2 - 9th EAI International Conference on IoT as a Service, IoTaaS 2023

Y2 - 27 October 2023 through 29 October 2023

ER -

Liu Y , Cui J, Long Z. A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift. In Chen X, Wang X, Lin S, Liu J, editors, IoT as a Service - 9th EAI International Conference, IoTaaS 2023, Proceedings. Springer Science and Business Media Deutschland GmbH. 2025. p. 183-199. (Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST). doi: 10.1007/978-3-031-70507-6_15

A Short Term Traffic Prediction Model Based on Deep Capture of Temporal Periodic Drift

Abstract

Publication series

Conference

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this