Twin-rotor Attitude Control using Adaptive Generalized Dynamic Inversion with Neural Estimation

This paper presents the Adaptive Generalized Dynamic Inversion control with Neural Estimation (AGDI-NE) for attitude control of affine nonlinear Twin Rotor Multiple-Input-Multiple-Output System (TRMS). The control problem is solved by exploiting the AGDI-NE control approach which is visualized as the combination of equivalent control and the switching control part. The equivalent control enforces the constraint dynamics that include the design objectives, where as the switching control provides robustness against system nonlinearities and uncertainties. The system unknown dynamical parameters are also estimated using Radial Basis Function Neural Networks (RBF-NN) for synthesizing the control law. The proposed control approach will guarantee semi-global practically stable attitude tracking in the sense of Lyapunov. Computer simulation are presented on the dynamic simulator of TRMS to demonstrate the effectiveness and feasibility of the AGDI-NE control law in presence of parametric uncertainties and measurement noise.