Feedback linearization with a neural network based compensation scheme

Abstract

This paper presents a nonlinear controller for uncertain single-input-single-output (SISO) nonlinear systems. The adopted approach is based on the feedback linearization strategy and enhanced by a Radial Basis Function neural network to cope with modeling inaccuracies and external disturbances that can arise. An application of this nonlinear controller to an electro-hydraulic actuated system subject to an unknown dead-zone input is also presented. The obtained numerical results demonstrate the improved control system performance. © 2012 Springer-Verlag.