Adaptive discrete-time control with dual neural networks for HFV via back-stepping

The article investigates the discrete-time controller for the longitudinal dynamics of the hypersonic flight vehicle. Based on the analysis of the control inputs, the dynamics model can be decomposed into the altitude subsystem and the velocity subsystem. Using the first-order Taylor expansion, the altitude subsystem can be transformed into discrete-time model, and then the strict-feedback form can be obtained. The controller is designed via back-stepping method. During this progress, neural networks are employed to approximate the mismatched uncertainties. Neural networks are used on the denominator of the controller as well as on the numerator of the controller to approximate the whole uncertainty (including the nominal value). The dual neural network controller via back-stepping is able to track system instructions accurately. Stability analysis proves that the errors of all the signals in the system are of uniform ultimate bound-ness. The simulation results show the effectiveness of the proposed controller.