EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Multi-Perspective Spatio-Temporal Feature Fusion Model for Urban Traffic Flow Prediction

Avazjon Marakhimov iD, Rustem Jalelov iD, Jabbar Kudaybergenov iD, Zahriddin Muminov iD, Kabul Khudaybergenov iD, Shukhrat Tajibaev

Sensors · 26.07.2026 · 26 (15) · 4744

DOI: 10.3390/s26154744 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 26.07.2026. Verfasst von Avazjon Marakhimov und 5 weiteren Autor:innen. Beteiligte Einrichtung: Department of Information Processing and Management Systems, Tashkent State Technical University, Tashkent 100174, Uzbekistan. Die Arbeit nennt 61 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Urban traffic flow is difficult to forecast accurately because its evolution is non-linear and governed by dependencies that operate over different spatial and temporal ranges. This paper introduces the Multi-Perspective Spatio-Temporal Feature Fusion Model (MPSTFFM) to describe these dependencies through complementary views. The temporal signal is separated into a slowly varying trend and a residual fluctuation, while the spatial structure is represented by four graphs: first-order adjacency, second-order in-degree, second-order out-degree, and a data-adaptive graph. These graphs respectively encode physical road connectivity, common inflow sources, common outflow destinations, and latent spatial associations. Whereas the first three are constructed from the known network topology, the adaptive graph is learned together with the prediction model and can therefore identify correlations not expressed by physical links. Within each spatio-temporal view, self-attention captures dependencies over long ranges, and convolutional operations extract local patterns. The features learned from all views are subsequently fused into a high-dimensional representation used to predict future flow. Experiments on real-world datasets compare MPSTFFM with twelve methods published during the preceding five years. On these benchmarks MPSTFFM outperforms every baseline, lowering the average MAE, RMSE, and MAPE across the four datasets by 13.04%, 5.28%, and 9.59%, respectively, relative to the best baseline on each one.

Schlagwörter

Stichwörter der Autor:innen

convolutional neural networksgraph neural networksself-attention mechanismspatio-temporal correlationtemporal fusiontraffic flow prediction

Autor:innen und Institutionen

  • Department of Information Processing and Management Systems, Tashkent State Technical University, Tashkent 100174, Uzbekistan.
  • Department of Software Engineering, Nukus State Technical University, Nukus 100147, Uzbekistan.
  • Department of Applied Mathematics, Tashkent State University of Economics, Tashkent 100066, Uzbekistan.
  • Department of Applied Informatics, Kimyo International University in Tashkent, Tashkent 100121, Uzbekistan.
  • Department of Exact Sciences, Mamun University, Urganch 220900, Uzbekistan.

ORCID-Profile: Avazjon Marakhimov, Rustem Jalelov, Jabbar Kudaybergenov, Zahriddin Muminov, Kabul Khudaybergenov

Zitieren

APA

Marakhimov, A., Jalelov, R., Kudaybergenov, J., Muminov, Z., Khudaybergenov, K., & Tajibaev, S. (2026). Multi-Perspective Spatio-Temporal Feature Fusion Model for Urban Traffic Flow Prediction. Sensors, 26(15), 4744. https://doi.org/10.3390/s26154744

Vancouver

Marakhimov A, Jalelov R, Kudaybergenov J, Muminov Z, Khudaybergenov K, Tajibaev S. Multi-Perspective Spatio-Temporal Feature Fusion Model for Urban Traffic Flow Prediction. Sensors. 2026;26(15):4744. doi:10.3390/s26154744

BibTeX
@article{Marakhimov2026,
  author = {Marakhimov, Avazjon and Jalelov, Rustem and Kudaybergenov, Jabbar and Muminov, Zahriddin and Khudaybergenov, Kabul and Tajibaev, Shukhrat},
  title = {Multi-Perspective Spatio-Temporal Feature Fusion Model for Urban Traffic Flow Prediction},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {15},
  pages = {4744},
  doi = {10.3390/s26154744}
}
Kostenlos starten →