EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

ATFM Delay Prediction With an Enhanced Index System: A Bayesian-Optimized XGBoost Model Incorporating Airway Network Characteristics

Yang Li, Zheng Zhao, Hang Zhou, Yanchun Li, Tianyang Li

IEEE Open Journal of Intelligent Transportation Systems · 2026 · 7 · 1034-1050

DOI: 10.1109/ojits.2026.3680722 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Open Journal of Intelligent Transportation Systems, 2026. Verfasst von Yang Li und 4 weiteren Autor:innen. Die Arbeit nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Air Traffic Flow Management (ATFM) delay quantitatively represents the spatiotemporal distribution of congestion arising from imbalances between capacity and demand within the airspace network. To enhance prediction accuracy, this study develops an enhanced index system that integrates airway-related factors encompassing structural and operational characteristics, together with critical delay determinants such as waypoint degree, average shortest path length, and congestion propagation rate. The proposed framework utilizes a Bayesian-optimized XGBoost (BO-XGBoost) model, wherein a Gaussian Process-guided search is implemented to identify an optimal hyperparameter configuration( $\text {n}_{\text {est}} \text = 269$ , $\text {d}_{\max }\text = 7$ ) that balances model complexity with generalization. Comparative benchmarking reveals that BO-XGBoost attains an accuracy of 81.0% and an Area Under the Curve (AUC) of 0.88, significantly outperforming Random Forest (AUC $\text = 0.85$ ) and LSTM (AUC $\text = 0.84$ ) baselines while exhibiting superior computational efficiency. Controlled ablation studies further substantiate the efficacy of the proposed index system. Compared to a baseline metric system excluding airway factors, the integrated approach yields improvements of 4% in accuracy, 9% in recall, and 4% in AUC. Relative to a system incorporating airway factors alone, the proposed model demonstrates gains of 7% in accuracy, 23% in recall, and 11% in AUC. Finally, SHAP-based interpretability analysis identifies airway flow dynamics and topological metrics as dominant predictors. These results underscore the reliability of the enhanced index system as a robust foundation for proactive ATFM decision-making in complex airspace environments.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Li, Y., Zhao, Z., Zhou, H., Li, Y., & Li, T. (2026). ATFM Delay Prediction With an Enhanced Index System: A Bayesian-Optimized XGBoost Model Incorporating Airway Network Characteristics. IEEE Open Journal of Intelligent Transportation Systems, 7, 1034-1050. https://doi.org/10.1109/ojits.2026.3680722

Vancouver

Li Y, Zhao Z, Zhou H, Li Y, Li T. ATFM Delay Prediction With an Enhanced Index System: A Bayesian-Optimized XGBoost Model Incorporating Airway Network Characteristics. IEEE Open Journal of Intelligent Transportation Systems. 2026;7:1034-1050. doi:10.1109/ojits.2026.3680722

BibTeX
@article{Li2026,
  author = {Li, Yang and Zhao, Zheng and Zhou, Hang and Li, Yanchun and Li, Tianyang},
  title = {ATFM Delay Prediction With an Enhanced Index System: A Bayesian-Optimized XGBoost Model Incorporating Airway Network Characteristics},
  journal = {IEEE Open Journal of Intelligent Transportation Systems},
  year = {2026},
  volume = {7},
  pages = {1034--1050},
  doi = {10.1109/ojits.2026.3680722}
}
Kostenlos starten →