Fault Detection and Classification for Sensor Faults of UAV by Deep Learning and Time-Frequency Analysis
Jing Huang, Mengna Li, Youmin Zhang, Lingxia Mu, Zihang Ao, Haihua Gong
Abstract
Sensor faults could occur in unmanned aerial vehicles (UAVs) during a mission of flight, which might deteriorate UAV’s performance or even cause catastrophe. Considering different types of sensor faults of a quadrotor UAV, a new fault detection and classification method based on time-frequency analysis (TFA) and deep learning (DL) technologies is proposed in this paper. Firstly, the data sets including different types of sensor fault in time-domain are generated randomly. The date sets are then transformed into time-frequency domain by short-time Fourier transform (STFT), resulting in time-frequency graph (TFG). Secondly, the time-frequency graph sets are used to train the deep network, by which the fault type can be rapidly classified with high accuracy. Finally, the simulations are carried out to verify the performance of the proposed algorithm and the accuracy reaches to 99.6%.