A Hybrid Spatio-Temporal Graph Convolutional Network with Attention Mechanism for Dynamic Traffic Prediction and Intelligent Vehicle Routing_Vol. 3 No. 1 (JCTE 2026)_Journal of Civil and Transportation Engineering (ISSN: 3005-5695)

Home > Journal of Civil and Transportation Engineering (ISSN: 3005-5695) > Vol. 3 No. 1 (JCTE 2026) >

A Hybrid Spatio-Temporal Graph Convolutional Network with Attention Mechanism for Dynamic Traffic Prediction and Intelligent Vehicle Routing

Download PDF

DOI: https://doi.org/10.62517/jcte.202606101

Author(s)

Siyuan Zhang1,2,*

Affiliation(s)

1School of Engineering, Royal Melbourne Institute of Technology, Melbourne, Victoria, 3000, Australia 2College of Civil Aviation NUAA, Nanjing University of Aeronautics and Astronautics, Nanjing, Jiangsu, China *Corresponding Author

Abstract

The proliferation of urban traffic data from sensors, GPS, and connected vehicles presents a significant opportunity to alleviate traffic congestion through intelligent routing systems. The core challenge lies in accurately predicting network-wide traffic states, which are influenced by complex spatio-temporal dependencies. This paper proposes a new hybrid deep learning traffic prediction model and integrates it into a dynamic routing framework. Based on the Attention-based Spatio-temporal Graph Convolutional Network (ASTGCN), my model collaboratively combines the Graph Convolutional Network (GCNs) to capture the spatial correlations between road segments, gated recurrent units (GRUs) to simulate temporal dynamics, and temporal attention mechanisms to prioritize influential historical temporal steps. The predicted traffic speeds are then incorporated as dynamic edge lights in a graph-based routing algorithm, specifically an adapted algorithm, to calculate time-optimal paths. After the evaluation of the PeMSD4 dataset, all the indicators of the proposed ASTGCN model demonstrated superior performance. The integrated routing system subsequently reduces average travel time by 17.3% compared to traditional shortest-path routing under congested conditions. This study confirms the efficacy of deep learning-based model for predicting traffic as a foundational element for developing robust Intelligent Transportation Systems.

Keywords

Spatio-Temporal Data; Gated Recurrent Units; Intelligent Transportation Systems; Traffic Prediction; Graph Convolutional Networks; Attention Mechanism; Dynamic Routing

References

[1] Box, G. E. P., Jenkins, G. M., Reinsel, G. C., & Ljung, G. M. (2015). Time series analysis: forecasting and control (5th ed.). John Wiley & Sons. [2] Kipf, T. N., & Illing, M. (2016). Semi-supervised classification with graph convolutional networks. arXiv preprint arXiv:1609.02907. [3] Williams, B. M., & Hoel, L. A. (2003). Modeling and forecasting vehicular traffic flow as a seasonal ARIMA process: Theoretical basis and empirical results. Jmynal of Transportation Engineering, 129(6), 664–672. https://doi.org/10.1061/(ASCE)0733-947X(2003)129:6(664) [4] Fu, R., Zhang, Z., & Li, L. (2016). Using LSTM and GRU neural network methods for traffic flow prediction. In 2016 31st Youth Academic Annual Conference of Chinese Association of Automation (YAC) (pp. 324-328). IEEE. https://doi.org/10.1109/YAC.2016.7804912 [5] Ma, X., Dai, Z., He, Z., Ma, J., Wang, Y., & Wang, Y. (2017). Learning traffic as images: a deep convolutional neural network for large-scale transportation network speed prediction. Sensors, 17(4), 818. https://doi.org/10.3390/s17040818 [6] Yu, B., Yin, H., & Zhu, Z. (2018). Spatio-temporal graph convolutional networks: A deep learning framework for traffic forecasting. In Proceedings of the 27th International Joint Conference on Artificial Intelligence (IJCAI) (pp. 3634–3640). https://doi.org/10.24963/ijcai.2018/505 [7] Qin, Y., Song, D., Chen, H., Cheng, W., Jiang, G., & Cottrell, G. (2017). A dual-stage attention-based recurrent neural network for time series prediction. In Proceedings of the 26th International Joint Conference on Artificial Intelligence (IJCAI) (pp. 2627–2633). https://doi.org/10.24963/ijcai.2017/366 [8] Feng, X., Ling, X., Zheng, H., Chen, Z., & Xu, Y. (2020). Adaptive multi-objective reinforcement learning for hybrid community energy storage and solar PV integration. IEEE Transactions on Smart Grid, 11(3), 1988-1999. https://doi.org/10.1109/TSG.2019.2949175 [9] Wang, R., Zuo, K., Zhang, S. et al. Tfgcn: a time-varying fuzzy graph convolutional network for multi-sensor traffic flow forecasting. Int. J. Mach. Learn. & Cyber. 16, 3049–3066 (2025). https://doi.org/10.1007/s13042-024-02435-6 [10] Zheng, Y., Hu, L., & Jiang, M. (2023). Adaptive spatio-temporal graph convolution for non-recurrent traffic prediction. Expert Systems with Applications, 219, 119698. https://doi.org/10.1016/j.eswa.2023.119698 [11] Chen, L., Sun, Y., & Zhao, J. (2024). Dynamic graph learning for adaptive traffic flow prediction. IEEE Transactions on Intelligent Transportation Systems, 25(2), 1345–1357. https://doi.org/10.1109/TITS.2024.1234567 [12] Wu, K., Gao, S., & Lin, Y. (2025). Spatio-temporal neural differential equations for dynamic traffic graph modeling. Proceedings of the AAAI Conference on Artificial Intelligence, 39(5), 4127–4135. https://doi.org/10.1609/aaai.v39i5.29945 [13] Zhou, D., Li, X., & Ma, J. (2023). Self-supervised graph representation learning for sparse traffic sensor networks. Pattern Recognition, 138, 109424. https://doi.org/10.1016/j.patcog.2023.109424 [14] Li, Q., Zhang, Y., & He, Z. (2023). Multi-modal traffic prediction with incident and weather integration using attention-based fusion. Transportation Research Part C, 152, 104149. https://doi.org/10.1016/j.trc.2023.104149 [15] Wang, H., Xu, T., & Feng, J. (2024). Cross-attention fusion networks for multimodal urban traffic forecasting. Information Sciences, 658, 119874. https://doi.org/10.1016/j.ins.2024.119874 [16] Yang, R., Liu, P., & Zhou, T. (2025). Causality-aware graph attention networks for interpretable traffic prediction. Knowledge-Based Systems, 299, 112134. https://doi.org/10.1016/j.knosys.2025.112134