LEARNING CONTROL DESIGN FOR A CLASS OF NONLINEAR-SYSTEMS

This paper proposes an adaptive method for the compensation of uncerta inties, for a class of nonlinear systems. A sliding mode control strat egy is used for the robust control design, using stable learning techn iques. Gaussian networks are used to identify the class of system unce rtainties. Learning and control bounds are guaranteed by properly cons tructing the training structure. The proposed technique has been valid ated using a hardware example case of an electromechanical robotic lin kage system. Experiments have shown that the inclusion of the proposed learning technique in the robust control design results in improved s ystem performance. (C) 1998 Elsevier Science Ltd. All rights reserved.