Robust actuator fault detection for quadrotor UAV with guaranteed sensitivity

Abstract

This paper proposes an actuator fault detection method for quadrotor UAV, in which sensitivity to fault and robustness to disturbance are simultaneously guaranteed. First, a quadrotor UAV system is converted to a linear-parameter-varying (LPV) form related to the altitude velocity and attitude angular velocity. Secondly, based on this LPV model, a fault detection method is designed, in which H− performance is applied to enhance fault sensitivity and L∞ performance is applied for guarantee robustness of the residual against the disturbance. And then the L∞ performance provides a detection threshold combined with initial flight condition for residual evaluation. Finally, based on the QDrone testbed, both numerical simulations and plentiful online experiments are implemented to validate the performance of the proposed fault detection scheme.