Improved adaptive NN backstepping control design for a perturbed PVTOL aircraft

Abstract

This paper proposes an improved flight control scheme for a perturbed planar vertical take-off and landing (PVTOL) aircraft based on backstepping and neural networks. In this approach, three neural networks are applied to deal with the external disturbances of the aircraft system. Instead of using σ-modification to design the adaptive laws, gradient descent algorithm is adopted to train the weight parameters of neural networks such that high function approximation precision can be achieved. By resorting to Lyapunov stability criterion, it can be proved that the reference signals are tracked by the aircraft’s outputs with small errors. In the end, simulation results are used to illustrate the superiority of the our approach compared with traditional direct adaptive neural network backstepping control technique.