Attitude Adaptive Neural Network Active Disturbance Rejection Control for Quad Tilt Rotor Unmanned Aerial Vehicle

The quad tilt rotor unmanned aerial vehicle (QTR-UAV) is a nonlinear tough coupled system with more redundant actuators. Its flight dynamics are affected by internal and external disturbances such as rotor tilt motion and wind disturbance. It is difficult to establish an accurate flight dynamics model and design a stable flight control system. In this paper, the flight dynamics model affected by internal and external disturbances is established based on the flight principle. The unknown disturbances are estimated through the adaptive neural network extended state observer. The redundant control strategy design is used to overcome the influence of rotor tilting. The rotor tilting process is stably controlled in the transition corridor as the flight speed changes. The channel control law algorithm uses a nonlinear state error feedback regulator to compensate for the influence of internal and external disturbances. The channel control outputs are distributed according to the manipulation strategy. The QTR-UAV flight safety and stability are ensured. A flight simulation system for the QTR is constructed, which is implemented with an adaptive radial basis function neural network active disturbance rejection control (RBF-ADRC). The attitude control simulation is conducted and some simulation results are presented. It is demonstrated that the designed RBF-ADRC flight control system can be effectively and reliably applied to the flight control of QTR, and its tracking accuracy, anti-interference and robustness are better than traditional active disturbance rejection controller and PID controller.