Potential Analysis of Current Technologies for UAV Rotor Blade Fault Diagnosis

Abstract

The demand for an automated fault detection system for UAVs is growing continuously in the face of the rising number of drones in commercial operation. In this project, a potential analysis of current technologies from different fields including acoustics, vibration analysis, electrical inspection, visual inspection, and flight data analysis is conducted. Special attention is given to the acoustic analysis as it is a promising and non-invasive method, which can be transferred to other systems. The findings of the literature review are evaluated in an experimental research. Test flights with a Holybro X500 UAV are performed and the logged data is used for a first basic approach to fault detection based on flight data. The acoustic signal of a single rotor is investigated in great detail with the use of a CAE Systems M112 acoustic camera. From the resulting acoustic images and Fourier analysis, major characteristic features of the rotor’s aeroacoustics and the influence of the noise of the brushless DC motor are identified and localized. It is confirmed that the main source of sound at frequencies roughly above 8000 Hz are the rotor blade tips. This is really important when considering blade tip damages as any changes in the acoustic signature of the blade tips should be easy to detect. Second, two rotor blades with different damages are compared to an undamaged blade. The difference is recognizable at higher frequencies, where the damaged rotors show a significantly higher amplitude in the frequency spectrum. This is verified by means of acoustic images. The acoustic signal is also compared to frequency data from an accelerometer. In the vibration signal no significant features can be found that would indicate a blade fault, except for a higher amplitude at some frequency peaks due to the mass imbalance. The acoustic signal proves to have a high potential for the development of a fault diagnosing system.