Online Mission-Profile-Aware Budgeted Anomaly Discovery for UAV Telemetry in Power-Inspection-Inspired Missions
IEEE Access · 2026 · 14 · 169848-169864
DOI: 10.1109/access.2026.3749380 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Mingxu Chen und 5 weiteren Autor:innen. Die Arbeit nennt 58 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Hongchuang Technology Branch under the 2025 project “Research on Network Security Testing and Protection Verification Technology Based on Transmission, Transfo….
Abstract (Originalsprache)
Unmanned aerial vehicles (UAVs) used for power-line inspection generate long telemetry streams, while the resources available for reviewing, annotating, and storing them are limited. We propose online mission-profile-aware budgeted anomaly discovery (O-MP-BAD), a method that selects telemetry windows under a fixed review budget. The method combines anomaly magnitude calibrated on normal data, boundary uncertainty, novelty, mission context, task-impact indicators, persistence, and redundancy control. A streaming top- $B$ buffer retains the selected evidence as windows arrive. Experiments use three public UAV datasets—UAV-SEAD, BASiC, and ALFA—and Profile C, a software-in-the-loop (SITL) simulation of a tower-inspection mission. For each dataset, one O-MP-BAD configuration and one classic online policy are selected on validation data and evaluated across six test budgets. O-MP-BAD increases the normalized area under the budget curve (AUC) for positive-sequence coverage by 0.085–0.251 on Profile C, UAV-SEAD, and BASiC, with recall-AUC changes between −0.005 and + 0.003. The classic policy achieves higher recall, precision, and sequence coverage on ALFA. A comparison using identical anomaly scores shows that the selection layer contributes to the coverage gains.
Förderung
- Hongchuang Technology Branch under the 2025 project “Research on Network Security Testing and Protection Verification Technology Based on Transmission, Transfo…
Zitieren
APA
Chen, M., Wang, X., Zhang, Q., Fu, X., Luo, H., & Qiu, T. (2026). Online Mission-Profile-Aware Budgeted Anomaly Discovery for UAV Telemetry in Power-Inspection-Inspired Missions. IEEE Access, 14, 169848-169864. https://doi.org/10.1109/access.2026.3749380
Vancouver
Chen M, Wang X, Zhang Q, Fu X, Luo H, Qiu T. Online Mission-Profile-Aware Budgeted Anomaly Discovery for UAV Telemetry in Power-Inspection-Inspired Missions. IEEE Access. 2026;14:169848-169864. doi:10.1109/access.2026.3749380
BibTeX
@article{Chen2026,
author = {Chen, Mingxu and Wang, Xueyan and Zhang, Qian and Fu, Xiaotian and Luo, Huaxi and Qiu, Tao},
title = {Online Mission-Profile-Aware Budgeted Anomaly Discovery for UAV Telemetry in Power-Inspection-Inspired Missions},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {169848--169864},
doi = {10.1109/access.2026.3749380}
}
EUVIMED