Adaptive neural network robust tracking design for a class of uncertain nonlinear system

This paper, based on radial basis function (RBF) neural network, presents an novel adaptive robust controller for a class of strict-feedback uncertainty nonlinear systems to address the tracking problem. The proposed approach, takes advantage of RBF neural network approximation property to approximate system uncertainties, and utilizes adaptive backstep-ping techniques for eliminating the effects of uncertainties with robust terms between actual controller and virtual controller. System adaptive laws, based on Lyapunov stability theory and RBF neural network weights matrix, are designed and derived, which can ensure all system signals are bounded, besides, the tracking error can converge to the neighborhood of zero given appropriate control parameters. This method does not require the upper bounds of the uncertainties of the system and their arbitrary order derivative. Simulation results illustrate the proposed method effectively.