Neural network-based model reference adaptive control system

摘要

In this paper, an approach to model reference adaptive controlnbased on neural networks is proposed and analyzed for a class ofnfirst-order continuous-time nonlinear dynamical systems. The controllernstructure can employ either a radial basis function network or anfeedforward neural network to compensate adaptively the nonlinearitiesnin the plant. A stable controller-parameter adjustment mechanism, whichnis determined using the Lyapunov theory, is constructed using anΣ-modification-type updating law. The evaluation of control errornin terms of the neural network learning error is performed. That is, thencontrol error converges asymptotically to a neighborhood of zero, whosensize is evaluated and depends on the approximation error of the neuralnnetwork. In the design and analysis of neural network-based controlnsystems, it is important to take into account the neural networknlearning error and its influence on the control error of the plant.nSimulation results showing the feasibility and performance of thenproposed approach are given