Continuous time nonlinear adaptive control based on linearization using neural network

The authors present a novel direct adaptive controller structure consisting of a neural net (NN) and a robust adaptive controller. The functional-link network (FLN) is utilized as the NN. The feature of the proposed scheme is that the FLN can compensate for the nonlinearity of the system to linearize the dynamics from the output of the adaptive controller to the system output, while the role of the adaptive controller is to perform the model-matching for the linearized system, which is referred to as the model reference adaptive control (MRAC). In this paper, we show some simple learning rules for adjusting the weights of the FLN. We will also give a new stability-guaranteed adaptive algorithm for adjusting the adaptive controller parameters and the weights of the FLN by treating the total system unifying the MRAC and the FLN. It will be shown that adequate cooperation of the NN with the MRAC will improve the convergence of the adaptation of the controller parameters and the weights of the FLN. Finally numerical simulation results will be given to examine the effectiveness of the proposed schemes.