Feedback linearization in discrete-time using neural networks

For a class of multi-input and multi-output nonlinear systems, multilayer neural network-based (NN) controllers which feedback linearize the nonlinear system are presented using two different methods. In the first method, one multilayer NN is employed whereas in the second method two multilayer NNs are used. For these two methods, novel weight tuning algorithms for the NN in discrete-time are developed, that are similar to ε-modification in the case of continuous-time adaptive control. The uniform ultimate boundedness of the tracking error and weight estimates is presented. No learning phase is needed for the NN and initialization of the network weights is straightforward. Simulation results justify the theoretical conclusions