An intelligent controller for underactuated mechanical systems

Abstract: This paper presents an intelligent controller for uncertain underactuated nonlinear systems. The adopted approach is based on sliding mode control and enhanced by an artificial neural network to cope with modeling inaccuracies and external disturbances that can arise. The sliding surfaces are defined as a linear combination of both actuated and unactuated variables. A radial basis function is added to compensate the performance drop when, in order to avoid the chattering phenomenon, the sign function is substituted by a saturation function in the conventional sliding mode controller. An application of the proposed scheme is introduced for an inverted pendulum, in order to illustrate the controller design method, and numerical results are presented to demonstrate the improved performance of the resulting intelligent controller.