Command-filtered-based adaptive neural control for nonlinear systems with actuator saturation

We report an anti-disturbance command-filtered-based adaptive control for nonlinear systems with actuator saturation. The adaptive neural control is developed based on a coordinate transform, High-order sliding mode(HOSM) observer and neural network (NN) approximation, while only one NN is employed to online approximate uncertain functions. To avoid potential windup caused by unknown input saturation, an NN compensator is added to perform as active disturbance rejection term in the feedforward manner. Moreover, a tracking differentiator(TD) is employed to avoid “explosion of complexity” in traditional backstepping method. Simulation results indicate that the derived scheme can effectively compensate for the disturbance caused by the saturation and system uncertainties.