Thrust-fault Diagnosis of Hexacopter UAV Using Supervised Learning With Disturbance Observers

This paper presents a real-time thrust fault diagnosis module for hexacopter UAVs, utilizing supervised learning and disturbance observers. The primary aim is to enhance the real-time diagnostic capabilities crucial for UAV safety and reliability. By employing disturbance observer technology, the proposed method effectively identifies and classifies thrust faults only using moment of Inertia data. The system was tested using GAZEBO simulations and real flight scenarios, demonstrating its effectiveness in accurately diagnosing faults. The research offers valuable insights into thrust fault diagnosis methodologies, contributing to improved fault-tolerant control systems for UAVs.