Neural robust adaptive hypersonic flight control without back-stepping

In this paper, one robust adaptive controller is proposed for hypersonic flight dynamics with high gain observer (HGO) and neural networks (NNs). For the attitude subsystem, it contains flight path angle, pitch angle and pitch rate. To avoid the complex back-stepping design, the dynamics are transformed into the normal feedback form. With HGO estimating the newly defined states, the controller is proposed with error feedback, NN approximation and the robust item. It is proved that the NN weights and filtered tracking error are guaranteed in the semiglobal sense. The simulation with March 15 is conducted to show the effectiveness of the proposed method.