Application of the Active Disturbance Rejection Control Based on Adaptive RBFNN Noise Estimating to Attitude Control

An attitude control method is designed for the rotor-flying platform to reject disturbances from the uncertainty of the intrinsic parameters of RF-MJM (rotor-flying multi-joint manipulator), the external environment, and the force that the manipulator reacts on the platform during the motion planning. Firstly, the internal and external disturbances of RF-MJM are asymptotically estimated by adaptive RBFNN (radial basis function neural network) algorithm and compensated in real time, setting the tracking differentiator (TD) as transient process of the desired attitude. Then, attitude tracking control of RF-MJM is accomplished by using nonlinear state error feedback (NLSEF) control, and the stability is analyzed by Lyapunov function. Finally, the algorithm is implemented on the simulation platform and its result is analyzed through comparing with PID (proportional-integral-differential) control and traditional ADRC (active disturbance rejection control). And the algorithm is testified in the realistic system of RF-MJM, where the three-axis attitude angles can be rapidly tracked from 0 to 0.6rad in 0.4s without overshoot. The proposed algorithm significantly outperforms ADRC and PID algorithms, for it is of strong anti-jamming capacity against disturbances from different channels, and of better robustness to system parameters. Results show that the proposed algorithm can effectively resolve the problem of uncertain disturbances in system, and track the attitude rapidly and accurately.