Multicopter Fault Detection and Identification via Data-Driven Statistical Learning Methods

What is it about?

Three data-driven methods for detecting and identifying rotor failures on a hexacopter has been demonstrated. One method requires knowledge of the system dynamics, whereas the other two do not. Statistical time-series models provide stochastic representations of the aircraft dynamics, from which fault-sensitive characteristic quantities and useful features may be derived to construct statistical hypothesis tests and train machine learning algorithms, respectively to achieve real-time rotor fault detection and identification.

Featured Image