Feedback linearization using neural networks

For a class of single-input, single-output (SISO), continuous-time nonlinear systems, a neural network-based controller is presented that feedback linearizes the system. Control action is used to achieve tracking performance for a state-feedback linearizable, but unknown nonlinear system. A global stability proof is given in the sense of Lyapunov. It is shown that all the signals in the closed-loop system and the control action are GUUB. No learning phase requirement is needed and initialisation of the network is straightforward.