With the acceleration of urbanization, traffic flow prediction has become a core problem in intelligent transportation systems. However, existing models generally overlook delay states, making it difficult to capture the dynamic evolution of traffic networks, and they also fall short in modeling the nonlinear relationships and spatial dependencies among nodes. To address these issues, this paper proposes a traffic flow prediction model named Delay-aware Adaptive Sparse Transformer (DASTFormer). Specifically, the model first employs a delay-enhanced feature fusion layer to capture propagation delays among nodes. Then, it dynamically updates the correlation matrix via mutual information and integrates it into the self-attention mechanism, thereby improving the modeling capability for complex spatiotemporal dependencies. Finally, an adaptive sparse spatial attention mechanism is introduced to focus on key nodes, enhancing prediction accuracy. Experiments conducted on two real-world traffic datasets demonstrate that DASTFormer outperforms mainstream Transformer-based baselines. Compared with the delay modeling method of the same type, PDFormer, DASTFormer reduces MAE and RMSE by 2.283% and 1.489%, respectively, on the PeMS08 dataset, validating the effectiveness of the proposed architecture.
WangQZhangKZhuC, et al. Traffic accident risk prediction for multi-factor spatio-temporal networks. IAENG Int J Appl Math2023; 53(4): 1315–1331.
2.
KumarNRaubalM. Applications of deep learning in congestion detection, prediction and alleviation: a survey. Transp Res Part C Emerg Technol2021; 133: 103432.
3.
SaleemiHRehmanZUKhanA, et al. Effectiveness of intelligent transportation system: case study of Lahore safe city. Transp Lett2022; 14(8): 898–908.
4.
WuYTanHQinL, et al. A hybrid deep learning based traffic flow prediction method and its understanding. Transp Res Part C Emerg Technol2018; 90: 166–180.
5.
YuHWuZWangS, et al. Spatiotemporal recurrent convolutional networks for traffic prediction in transportation networks. Sensors2017; 17(7): 1501.
6.
MaXDaiZHeZ, et al. Learning traffic as images: a deep convolutional neural network for large-scale transportation network speed prediction. Sensors2017; 17(4): 818.
7.
ZhangJZhengYQiD, et al. Predicting citywide crowd flows using deep spatio-temporal residual networks. Artif Intell2018; 259: 147–166.
8.
WengWFanJWuH, et al. A Decomposition dynamic graph convolutional recurrent network for traffic forecasting. Pattern Recognit2023; 142: 109670.
9.
LiYYuRShahabiC, et al. Diffusion convolutional recurrent neural network: data-driven traffic forecasting. In: International conference on learning representations, Vancouver, BC, Canada, 2018.
10.
HeTLiuYOngY-S, et al. Polarized message-passing in graph neural networks. Artif Intell2024; 331: 104129.
11.
ZuoJZeitouniKTaherY, et al. Graph convolutional networks for traffic forecasting with missing values. Data Min Knowl Discov2023; 37(2): 913–947.
12.
ShenGZhouWZhangW, et al. Bidirectional spatial–temporal traffic data imputation via graph attention recurrent neural network. Neurocomputing2023; 531: 151–162.
13.
ShiMTangYZhuX, et al. Genetic-GNN: evolutionary architecture search for graph neural networks. Knowl Based Syst2022; 247: 108752.
GuoKHuYQianZS, et al. An optimized temporal-spatial gated graph convolution network for traffic forecasting. IEEE Intell Transp Syst Mag2022; 14(1): 153–162.
16.
SongCLinYGuoS, et al. Spatial-temporal synchronous graph convolutional networks: a new framework for spatial-temporal network data forecasting. Proc AAAI Conf Artif Intell2020; 34: 914–921.
17.
ChenYLiKYeoCK, et al. Traffic forecasting with graph spatial–temporal position recurrent network. Neural Netw2023; 162: 340–349.
18.
LongQFangZFangC, et al. Unveiling delay effects in traffic forecasting: a perspective from spatial-temporal delay differential equations. In: Proceedings of the ACM web conference 2024, Singapore, 13–17 May 2024, pp. 1035–1044. New York: ACM.
19.
TangJYangJZhangY. D-STANet: a delay-enhanced spatio-temporal attention network for traffic prediction. Knowl Based Syst2025; 320: 113533.
20.
JiangJHanCZhaoWX, et al. Pdformer: propagation delay-aware dynamic long-range transformer for traffic flow prediction. Proc AAAI Conf Artif Intell2023; 37(4): 4365–4373.
21.
LiuGWuYZhaoD, et al. Time-adaptive graph convolutional network for traffic prediction. In: Proceedings of the 2021 5th international conference on deep learning technologies, Qingdao, China, 23–25 July 2021, pp. 81–86. New York: ACM.
22.
MaQSunWGaoJ, et al. Spatio-temporal adaptive graph convolutional networks for traffic flow forecasting. IET Intell Transp Syst2023; 17(4): 691–703.
23.
LiuAZhangY. Spatial–temporal dynamic graph convolutional network with interactive learning for traffic forecasting. IEEE Trans Intell Transp Syst2024; 25(7): 7645–7660.
24.
SemenoglouA-ASpiliotisEAssimakopoulosV. Image-based time series forecasting: a deep convolutional neural network approach. Neural Netw2023; 157: 39–53.
25.
BiJZhangXYuanH, et al. A hybrid prediction method for realistic network traffic with temporal convolutional network and LSTM. IEEE Trans Autom Sci Eng2021; 19: 1869–1879.
26.
FuZFuZLiuQ, et al. SparseTT: visual tracking with sparse transformers. arXiv preprint arXiv:2205.03776, 2022.
27.
ZhangQLiCSuF, et al. Spatiotemporal residual graph attention network for traffic flow forecasting. IEEE Internet Things J2023; 10(13): 11518–11532.
28.
MazaréPEDehwahAHClaudelCG, et al. Analytical and grid-free solutions to the Lighthill–Whitham–Richards traffic flow model. Transp Res Part B Methodol2011; 45(10): 1727–1748.
YuBYinHZhuZ. Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting. arXiv preprint arXiv:1709.04875, 2017.
31.
ZhuJWangQTaoC, et al. AST-GCN: attribute-augmented spatiotemporal graph convolutional network for traffic forecasting. IEEE Access2021; 9: 35973–35983.
32.
FangZLongQSongG, et al. Spatial-temporal graph ODE networks for traffic flow forecasting. In: Proceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining, Singapore, 14–18 August 2021, pp. 364–373. New York: ACM.
33.
PuBLiuJKangY, et al. MVSTT: a multiview spatial-temporal transformer network for traffic-flow forecasting. IEEE Trans Cybern2022; 54(3): 1582–1595.
34.
WuZPanSLongG, et al. Graph WaveNet for deep spatial-temporal graph modeling. arXiv preprint arXiv: 1906.00121, 2019.
35.
BaiLYaoLLiC, et al. Adaptive graph convolutional recurrent network for traffic forecasting. Adv Neural Inf Process Syst2020; 33: 17804–17815.
36.
WuZPanSLongG, et al. Connecting the dots: multivariate time series forecasting with graph neural networks. In: Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, Virtual Event, CA, USA, 6–10 July 2020, pp. 753–763. New York: IEEE.
37.
ZhengCFanXWangC, et al. GMAN: a graph multi-attention network for traffic prediction. Proc AAAI Conf Artif Intell2020; 34(1): 1234–1241.
