DO-based dynamic neural network identification and anti-disturbance control with asymmetrical dead-zone constraints

This paper is concerned with the problem of neural network identification and anti-disturbance control of a class of complex nonlinear systems with unknown exogenous disturbances and asymmetrical dead-zone constraints. First, together with a disturbance observer (DO) which is designed to estimate unknown exogenous disturbances, the dynamic neural network (DNN) identifier is used to approximate the complex nonlinear systems. It is shown that both the identification errors of dynamic neural networks and the estimation errors of the disturbance observer can converge to zero. Moreover, a new disturbance observer based feedback controller is designed with the Nussbaum gain matrix so as to guarantee the designed DNN identifier to achieve a satisfactory anti-disturbance control performance. Finally, the applicability of the proposed algorithm is validated with simulation results.