Fault Detection and Diagnosis Techniques for Electric UAV Powertrain System

Abstract

Current growth of unmanned aerial vehicles (UAVs) suggests heavier low altitude traffic in urban airspace in the nearby future. UAV application areas include small package delivery drones, larger on demand mobility vehicles, and emergency fire suppression drones among numerous others. Safe and reliable operations of such vehicles become a more critical issue in low altitude and high number of takeoff/landing situations. In such applications, fast and accurate fault detection and isolation are very crucial safety and mission concerns. Efficient health monitoring also facilitates real-time decision support and repair/return-to-fleet decisions. Powertrains are vital subsystems of any electric UAV. Failures in any component or module of a powertrain may cause a mission failure, safety issues, and overall reliability concern. Moreover, timely information about failure in the powertrain helps to make decisions on whether to abort the mission or complete the mission with some functions limited (limp mode). This work presents a framework for fault detection and diagnosis techniques that are suitable for powertrain failures in UAV systems. For certain faults, signal based detection techniques are fast and reliable whereas for certain other faults, parameter estimation algorithms may be more suitable for better accuracy. In this work, we present an integrated model-based and data driven approach of fault detection and fault isolation at two different levels.