Neural networks for modeling and control of dynamic systems: a practitioner's handbook: M. Nørgaard, O. Ravn, N.K. Poulsen, and L.K. Hansen; Springer, London, 2000, 246pp., paperback, ISBN 1-85233-227-1

& Ersue, 1992) published a few years back on the subject of neurocontrol. The subject of neural networks for modeling and control of dynamical systems represents an important area of applications of artiÿcial neural networks that reemerged in 1980s. On the other hand, to control engineers , neurocontrol, or neural networks for controls, brings about some new and fresh ideas to the modeling and control design for nonlinear dynamical systems. The approach based on neural networks is generally viewed as an important milestone in identiÿcation and adaptive control of nonlinear dynamical systems. Neural Networks for Modeling and Control of Dynamic Systems deals with control problems of unknown nonlinear dynamical systems using neural networks. One of the chapters (Chapter 2) deals with system identiÿcation of nonlin-ear dynamical systems using neural networks. The next two chapters (Chapters 3 and 4) deal with the control of non-linear dynamical systems using neural networks and several case studies. With an introductory chapter at the beginning, the book has only four chapters. Chapter 1 introduces the background information about neural networks. In particular, it introduces in detail the feedforward neural network structure, which is termed as multilayer perceptron networks in this book. The book is extended from a Ph.D. thesis done by the ÿrst author of the book. It is thus not a surprise that the book does not cover various neural network structures studied in the literature. The main object used in the book in terms of neural networks is the two-layer feedforward neural networks with sigmoidal type activation function for the hidden layer and linear function for the output layer. Other popular structures used often by control engineers such as radial basis function networks and recurrent neural networks are not mentioned at all since they are not required in the book later on. Chapter 2 provides details about how to use neural networks (i.e., feedforward neural networks or multilayer perceptron networks) for the purpose of nonlinear system identiÿcation. It covers model structure selection, data collection, neural network training (weight determination), and validation of the training. It presents a very clear picture for the purpose of nonlinear system modeling and identiÿcation using neural networks. To this reviewer, the whole chapter seems to be a very good summary of what is published and known in the literature on the subject of nonlinear system identiÿcation using neural networks. It is noted that the dynamical systems studied in …