Neural-network-based incremental backstepping sliding mode control for flying-wing aircraft

Abstract

The nonlinear trajectory tracking control problem is studied for a flying-wing aircraft. Starting from a nonlinear dynamics model of the flying-wing aircraft, the trajectory tracking control is decomposed into multiple loops of position control, flight path control, and attitude control. An incremental backstepping sliding mode control is proposed to implement attitude control, while an incremental nonlinear dynamic inversion and a nonlinear dynamic inversion design are used to deal with the nonlinear system model for the flight path and position control, respectively. In addition, a radial basis function neural-network-based extended state disturbance observer is proposed to deal with model uncertainties, gust disturbances, and unknown faults of the aircraft. The closed-loop control system is proved to be stable using Lyapunov theory. The performance of the proposed disturbance-observer-based incremental backstepping sliding mode control is demonstrated in simulation through a set of three-dimensional tracking scenarios. Compared with both backstepping control and backstepping sliding mode control, tracking performance measured by settling time, tracking error, and overshoot are improved by the proposed design when realistic trajectory tracking missions are executed.