Janczak, Andrzej Identification of Wiener and Hammerstein systems with neural networks and polynomials models. Methods and applications. (English) Zbl 1027.93002 Monographs. 2. Zielona Góra: University of Zielona Góra Press, xv, 185 p. (2003). The monograph deals with the identification of discrete-time Wiener and Hammerstein systems in a stochastic environment, using two types of models of nonlinear static characteristics: neural network models and polynomial models. Chapter 1 gives a brief review of the existing identification methods of the systems under consideration and presents standard model structures of discrete-time dynamical systems. Chapters 2 and 3 concern the neural network models of Wiener and Hammerstein systems, respectively, and provide algorithms for the calculation of the gradient (or approximate gradient) of the neural model output w.r.t. model parameters. Some new ideas concerning model parameter adjusting with gradient-based techniques are presented here. In turn, Chapters 4 and 5 are devoted to the presentation of the Wiener and Hammerstein identification routines based on polynomial models of nonlinearities. In particular, least squares and combined least squares-instrumental variables approaches are applied to the estimation of the linear dynamic subsystem parameters and the parameters of the nonlinear characteristic or its inverse. The applications of Wiener and Hammerstein models in nonlinear system modelling, control and fault diagnostics given in Chapter 6 complete the book. The approximation accuracy, computational complexity and advantages and disadvantages of the methods are discussed and compared in a unified framework. Reviewer: Zygmunt Hasiewicz (Wrocław) Cited in 1 Document MSC: 93-02 Research exposition (monographs, survey articles) pertaining to systems and control theory 93B30 System identification 93C10 Nonlinear systems in control theory 93C55 Discrete-time control/observation systems 93E12 Identification in stochastic control theory 93E24 Least squares and related methods for stochastic control systems 92B20 Neural networks for/in biological studies, artificial life and related topics Keywords:discrete-time Wiener systems; identification; neural network models; polynomial models; Hammerstein systems; least squares-instrumental variables approaches; modelling; fault diagnostics; approximation accuracy; computational complexity PDFBibTeX XMLCite \textit{A. Janczak}, Identification of Wiener and Hammerstein systems with neural networks and polynomials models. Methods and applications. Zielona Góra: University of Zielona Góra Press (2003; Zbl 1027.93002)