Neural Network Adaptive Control for a Class of Matched SISO Nonlinear Uncertain Systems with Zero Dynamics

The paper presents a direct adaptive tracking control scheme for a class of matched SISO affine nonlinear uncertain systems with zero dynamic using neural network. Through neural network approximation, neural network is used as the emulator of the unknown ideal controller. A quadratic cost function of the error between the unknown ideal controller and the used neural network controller is minimized using a gradient descent method to adjust parameters in neural network. The convergence of parameters and the uniformly ultimately boundedness of tracking error and all states of the closed-loop system are guaranteed based on Lyapunov stability theorem. The effectiveness of the proposed controller is illustrated through the simulation results.