Stable direct adaptive neurocontrol of nonlinear system

This paper investigates neural adaptive control of an unknown nonlinear system. The system is feedback linearizable. The control law is implemented by a neural network to solve the asymptotic tracking problem. Unlike other approaches, stability of the complete system is explicitly taken in account in the design of the neural net learning law. As a result adaptive asymptotic tracking is realized which guarantee tracking error convergence and local stability for the overall system. Experimental results illustrate this approach