Neural Network-Based Adaptive Sliding Mode Control Strategy for Underactuated RTAC

For the stabilization problem of underactuated RTAC (Rotational/Translational ACtuator), a sliding mode control (SMC) strategy is designed and presented based on neural network and active disturbance compensation. To overcome the underactuated characteristics, a new "actuated" state is constructed as an output by combining the underactuated and actuated states to transform the underactuated model to an actuated form, and the linear extended state observer (LESO) is employed for the total disturbance of the proposed model. By applying the neural network to approximate the optimal control outputs, the proposed method can achieve satisfactory control performance without a precise dynamic model. Furthermore, The stability analysis of the RTAC is presented based on rigorous Lyapunov method. The simulations are provided to validate the method. The simulations with the existing method are also provided as a comparison.