Neuro control of nonlinear discrete time systems with deadzone and input constraints

A neural network (NN) controller in discrete time is designed to deliver a desired tracking performance for a class of uncertain nonlinear systems with unknown deadzones and magnitude constraints on the input. The NN controller consists of two NNs: the first NN for compensating the unknown deadzones; and the second NN for compensating the uncertain nonlinear system dynamics. The magnitude constraints on the input are modeled as saturation nonlinearities and they are dealt with in the Lyapunov-based controller design. The uniformly ultimate boundedness (UUB) of the closed-loop tracking errors and the neural network weights estimation errors is demonstrated via Lyapunov stability analysis.