EUVIMEDEUVIMED entdecken
Crossref

EFformer: Exogenous Variable Fusional Transformer for Intersection-Level Turning Movement Traffic Flow Prediction

Yancheng Gong, Yangming Xu, Gongming Wang, Guangyuan Pan, Hao Wei

IEEE Systems, Man, and Cybernetics Magazine · 2026-07 · 12 (3) · 80-87

DOI: 10.1109/msmc.2025.3600845 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Systems, Man, and Cybernetics Magazine, 2026-07. Verfasst von Yancheng Gong und 4 weiteren Autor:innen. Die Arbeit nennt 22 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Intersection-level turning movement traffic flow prediction is critical for fine-grained traffic signal control and congestion management. Yet, existing approaches struggle in solving this due to the large number of predictive variables and the unclear intricate coupling relationships among them. In this article, we propose a hypothesis that exogenous factors (particularly road infrastructure characteristics and weather conditions) play a critical role in intersection-level turning movement traffic flow prediction. Based on this, we introduce the Exogenous Variable Fusional Transformer (EFformer) as a potential solution. In this model, the relevant exogenous variables through a specially designed variable fusion layer are integrated. Then, a spatiotemporal encoder for modeling spatiotemporal dependencies is proposed. In addition, a single-channel Mamba mechanism and a cross-dimension cross-attention mechanism are further introduced to highlight critical features across different dimensions. Finally, experiments are conducted using over 10 million real-world records from more than 100 intersections across six cities in Ontario, Canada, spanning nearly three years. The results show EFformer significantly outperforms other existing state-of-the-art (SOTA) models. Moreover, as a time series prediction framework that leverages exogenous variables, the findings of this article can also serve as a reference and provide valuable insights for similar research in other fields.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Gong, Y., Xu, Y., Wang, G., Pan, G., & Wei, H. (2026). EFformer: Exogenous Variable Fusional Transformer for Intersection-Level Turning Movement Traffic Flow Prediction. IEEE Systems, Man, and Cybernetics Magazine, 12(3), 80-87. https://doi.org/10.1109/msmc.2025.3600845

Vancouver

Gong Y, Xu Y, Wang G, Pan G, Wei H. EFformer: Exogenous Variable Fusional Transformer for Intersection-Level Turning Movement Traffic Flow Prediction. IEEE Systems, Man, and Cybernetics Magazine. 2026;12(3):80-87. doi:10.1109/msmc.2025.3600845

BibTeX
@article{Gong2026,
  author = {Gong, Yancheng and Xu, Yangming and Wang, Gongming and Pan, Guangyuan and Wei, Hao},
  title = {EFformer: Exogenous Variable Fusional Transformer for Intersection-Level Turning Movement Traffic Flow Prediction},
  journal = {IEEE Systems, Man, and Cybernetics Magazine},
  year = {2026},
  volume = {12},
  number = {3},
  pages = {80--87},
  doi = {10.1109/msmc.2025.3600845}
}
Kostenlos starten →