Fault diagnosis and condition monitoring of UAV rotor using signal processing

In this paper, a method for fault detection of physical impairment of UAV rotor blades is presented. Actuators in multirotor UAV (Unmanned Aerial Vehicle) systems are common subjects fault diagnosis methods which are an essential part of the active fault-tolerant control scheme. Defects in a propulsion system of the aerial vehicle lead to the loss of thrust generated by rotors and as a result, to the disturbance of thrust balance, higher power consumption and further degradation resulting in the possible crash of the vehicle. Authors propose a three-stage algorithm based on the signal processing and machine learning to detect the occurrence of rotor fault, determine its scale and type. The method is based on measurements of acceleration from the onboard IMU (Inertial Measurement Unit) sensor as unbalanced rotating parts commonly cause vibrations in mechanical systems. The acceleration signal is stored in a cyclic buffer and then processed by simple feature extraction algorithms in order to obtain a characteristic signature of the faulty state. Three different methods of feature extraction are considered in this article, along with the analysis of variable buffer length. Next, the Support Vector Machine (SVM) classifier is used to determine the occurrence and character of the rotor fault. The presented solution was verified in series of experiments proving its effectiveness. In addition, such approach based on signal processing is very versatile and easy to implement in arbitrary flight controller.