Adaptive single neural network control for a class of uncertain discrete-time nonlinear strict-feedback systems with input saturation

In this paper, a novel adaptive neural network control approach is presented for a class of uncertain discrete-time nonlinear strict-feedback systems with input saturation. By combining single neural network approximation and minimal learning parameter technique, the proposed approach is able to eliminate the complexity growing problem and alleviate the explosion of learning parameters. An auxiliary design system is incorporated into the control scheme to overcome the problem of input saturation constraints. Following this approach, the designed controller contains only one actual control law and one adaptive law, the numbers of input variables and weights of neural network updated online are decreased drastically, and the number of parameter updated online for whole system is reduced to only one. Compared with the existing methods, the adaptive mechanism with much simpler controller structure and minimal learning parameterization is achieved; therefore, the computational burden is lighter. It is shown via Lyapunov theory that all signals in the closed-loop system are uniformly ultimately bounded. Finally, simulation results via two examples are employed to illustrate the effectiveness and merits of the proposed scheme.