AEDA-Net: Attention-Embedded Domain Adaptation Network for UAV Fault Detection Under Heterogeneous Flight Status
IEEE Transactions on Instrumentation and Measurement · 2026 · 75 · 3518511-3518511
DOI: 10.1109/tim.2026.3706227 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Instrumentation and Measurement, 2026. Verfasst von Zhuocheng Cai und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 35 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Sichuan Science and Technology Program, Open Fund of Intelligent Control Laboratory.
Abstract (Originalsprache)
With the increasing deployment of unmanned aerial vehicles (UAVs) in safety-critical missions, ensuring reliable fault detection (FD) under heterogeneous flight status has become essential. Traditional methods typically rely on global modeling or domain-specific training, which often fail to adapt to variations in flight status and cannot effectively handle domain shifts. Although several improved frameworks have been proposed, they still struggle to capture status-dependent features and maintain robustness across diverse flight statuses. To address these challenges, this article proposes an attention-embedded domain adaptation network (AEDA-Net) for UAV FD under heterogeneous flight status. AEDA-Net augments the domain-adversarial learning paradigm by incorporating an attention-enhanced feature extractor that adaptively recalibrates sensor-level and status-specific representations before domain alignment. The embedded attention modules highlight fault-sensitive information while suppressing redundant variations induced by different flight statuses, thereby enhancing both discriminability and transferability. A gradient-reversal layer (GRL) and a domain discriminator are further employed to achieve stable and robust feature alignment between simulation data and real-flight data. Experimental evaluations using both simulation data and real-flight measurements demonstrate that AEDA-Net achieves high detection accuracy, strong generalization capability, and consistently strong performance when transferred across different flight statuses.
Förderung
- National Natural Science Foundation of China
- Sichuan Science and Technology Program
- Open Fund of Intelligent Control Laboratory
- System of Systems and Artificial Intelligence Key Laboratory of Sichuan Province
- Open Research Project of the State Key Laboratory of Industrial Control Technology, China
Zitieren
APA
Cai, Z., Zhang, Y., & Miao, Q. (2026). AEDA-Net: Attention-Embedded Domain Adaptation Network for UAV Fault Detection Under Heterogeneous Flight Status. IEEE Transactions on Instrumentation and Measurement, 75, 3518511-3518511. https://doi.org/10.1109/tim.2026.3706227
Vancouver
Cai Z, Zhang Y, Miao Q. AEDA-Net: Attention-Embedded Domain Adaptation Network for UAV Fault Detection Under Heterogeneous Flight Status. IEEE Transactions on Instrumentation and Measurement. 2026;75:3518511-3518511. doi:10.1109/tim.2026.3706227
BibTeX
@article{Cai2026,
author = {Cai, Zhuocheng and Zhang, Yujie and Miao, Qiang},
title = {AEDA-Net: Attention-Embedded Domain Adaptation Network for UAV Fault Detection Under Heterogeneous Flight Status},
journal = {IEEE Transactions on Instrumentation and Measurement},
year = {2026},
volume = {75},
pages = {3518511--3518511},
doi = {10.1109/tim.2026.3706227}
}
EUVIMED