Adaptive Control Using Neural Networks and Approximate Models 人工神经网络理论及应用 教学课件.pdfVIP

IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 8, NO. 3, MAY 1997 475 Adaptive Control Using Neural Networks and Approximate Models Kumpati S. Narendra, Fellow, IEEE, and Snehasis Mukhopadhyay Abstract—The NARMA model is an exact representation of the in practical contexts, have determined their acceptance by the input–output behavior of finite-dimensional nonlinear discrete- scientific community. time dynamical systems in a neighborhood of the equilibrium From the very beginning, it has been realized by systems state. However, it is not convenient for purposes of adaptive control using neural networks due to its nonlinear dependence theorists that most real dynamical systems are nonlinear. How- on the control input. Hence, quite often, approximate methods ever, linearizations of such systems around the equilibrium are used for realizing the neural controllers to overcome com- states yield linear models which are mathematically tractable. putational complexity. In this paper, we introduce two classes In particular, based on the superposition principle, the output of models which are approximations to the NARMA model, and of the system can be computed for any arbitrary input, and which are linear in the control input. The latter fact substantially simplifies both the theoretical analysis as well as the practical im- alternately, in control problems, the input which optimizes the plementation of the controller. Extensive simulation studies have output in some sense can also be determined with relative ease. shown that the neural controllers designed using the proposed