Neural Network-Based DOBC for a Class of Nonlinear Systems With Unmatched Disturbances

In this brief, the problem of composite anti-disturbance tracking control for a class of strict-feedback systems with unmatched unknown nonlinear functions and external disturbances is investigated. A disturbance-observer-based control (DOBC) in combination with a neural network scheme and back-stepping method is developed to achieve a composite anti-disturbance controller design that provides guaranteed performance. In the proposed method, a conventional disturbance observer and a radial basis function neural network (RBFNN) are combined into a new disturbance observer to estimate the unmatched disturbances. As compared with conventional DOBC methods, the primary merit of the proposed method is that the unknown nonlinear functions are approximated using the RBFNN technique, and not regarded as part of the disturbances or estimated by a conventional disturbance observer. Hence, the proposed method can obtain higher control accuracy than the conventional DOBC methods. This advantage is validated by simulation studies.