Peaking Free HGO Based Neural Hypersonic Flight Vehicle Control

Abstract

This paper describes the design of adaptive neural controller for the longitudinal dynamics of a generic hypersonic flight vehicle (HFV). For the altitude subsystem, the dynamics are transformed into the normal feedback form and the high gain observer (HGO) is taken to estimate the unknown newly defined states. Only one Neural Network (NN) is employed to approximate the lumped uncertain system nonlinearity which is considerably simpler than the back-stepping scheme with the strict-feedback form. Furthermore, the saturation design is applied on the HGO estimation error to eliminate the peaking phenomenon. For the velocity subsystem, dynamic inverse NN controller is designed. The Lyapunov stability of the NN weights and filtered tracking error are guaranteed in the semi global sense. The effectiveness of the proposed strategy is verified by numerical simulation study.