UAV fault detection improves with graph learning and language models
TGTV: An Interpretable Framework for UAV Fault Detection and Traceability via Spatiotemporal Graph Learning and Large Language Models
Machine Learning
Summary
Drones sometimes have hidden problems that are hard to spot because their sensors and controls behave in complex ways. The authors propose a new system called TGTV that uses advanced methods to watch how sensor readings change over time and how they relate to each other. TGTV combines this sensor data analysis with a language model that explains what went wrong in simple terms. This helps detect faults more accurately and provides clear evidence for understanding and fixing issues after a flight.
What this means in practice
- •For drone operation teams: Detect and explain faults in fixed-wing UAVs after flight using combined sensor data and language model analysis for clearer maintenance decisions.
- •For industrial iot maintenance teams: Identify and trace sensor faults in complex equipment by using spatiotemporal graphs and interpretable anomaly diagnosis similar to drone fault systems.
Authors
Zixuan Wang, Changbo Shi, Ji Liu
Abstract
Fixed-wing unmanned aerial vehicles experience operating-condition transitions, asynchronous sensor sampling, and control-system compensation that can obscure fault signatures. Existing temporal detection methods have limited capacity to represent coupled faults across subsystems, track their evolution, and constrain language-model diagnoses with quantitative evidence. This paper proposes TGTV, an interpretable fault detection and traceability framework that integrates spatiotemporal graph learning with a large language model. A temporal convolutional network and a graph attention network with weighted fused adjacency extract multiscale temporal patterns and spatial dependencies among sensors. A hybrid temporal Transformer and variational autoencoder then evaluates window-level anomalies using prediction errors and reconstruction probabilities. Continuous-time Shapley attribution identifies critical sensors and high-contribution intervals while preserving the temporal continuity of fault evolution. A structured evidence package combines detection results, graph relationships, temporal attribution, sensor attribution, and operating context. An evidence-constrained diagnosis protocol uses this package to generate flight-level conclusions with traceable evidence identifiers. Cross-flight holdout experiments on the public ALFA fixed-wing flight dataset yield a window-level area under the precision-recall curve of 0.976, accuracy of 0.973, recall of 0.902, and F1 score of 0.920. Flight-level fault diagnosis achieves recall of 0.943 and an F1 score of 0.944. These results demonstrate accurate anomaly detection and quantitatively supported fault-tracing cues, providing a basis for interpretable post-flight fault review and analysis.