Adaptive robust neural tracking control of a class of unknown nonlinear systems

In this paper, the adaptive robust output tracking for a class of nonlinear systems with unknown nonlinearities is concerned. Based on input-output linearization technique and the sliding control strategy, a neural-network-based adaptive robust control law is developed. The technique of the multilayered neural networks is employed to perform approximating input-output linearization. The adaptive robust sliding control is used to compensate the neural network approximation errors so that no prior knowledge of the bound on the approximation errors is required. The parameters of sliding control and the weights of the neural networks are updated according to the Lyapunov principle. It is shown that the outputs of the closed-loop system asymptotically track the desired output trajectories despite the neural network approximation errors, and the tracking errors can be made arbitrarily small while maintaining the boundedness of all signals within the system. The effectiveness and robustness of the proposed control scheme are demonstrated in the case of two-degree-of-freedom robotic manipulator.