Model-based Fault Detection and Identification of a Quadrotor with Rotor Fault

Fault detection and identification (FDI) is a challenging and critical process when dealing with nonlinear, unstable, and underactuated systems such as multirotor. This article presents a novel two-stage structure for a fault-tolerant FDI approach for a quadrotor with an actuator fault. The FDI algorithm generates residual signals for fault detection using a model-based approach based on the parity space. The basic idea behind this approach is to leverage measurement coherence by generating residuals via linear combinations of measurement outputs and control inputs over a finite window. The parity vector is generated using the states of the faulty system, which have been filtered using an extended Kalman filter, the inputs, and the healthy quadrotor model. To detect and identify the actuator's partial fault, the residual signal is examined using the exponential forgetting factor recursive least square method. Real-time testbed experiments are used to determine the FDI algorithm's performance and to demonstrate the proposed algorithm's effectiveness in identifying a quadrotor's rotor fault.