Fuzzy logic, identification and predictive control. (Q1762577)

New insights into the transfer of fuzzy methods into the modern control paradigms encompassing robust, model-based, PID-like, and predictive control are presented in this book. The presentation is balanced between fuzzy models (Ch.~1--4) and fuzzy control (Ch.~5--8). The approximation capabilities of triangular, polynomial, and Gaussian types of membership functions are analyzed. It is shown that any continuous function can be approximated with an arbitrary accuracy by a fuzzy inference system on a compact domain (Ch.~1). Main techniques to construct fuzzy models from input-output data are recalled and compared. The problems of generalization and consequence estimation are outlined. It is stated that the fuzzy model will be at least as good as the best multi-linear approximation (Ch.~2). The concept of linguistic integrity is introduced, and an algorithm to build up fuzzy models with linguistic meaning is proposed. Fuzzy models of complex nonlinear functions, chaotic time series, and industrial processes are elaborated using this model-building tool (Ch.~3). A framework for identification of nonlinear dynamic systems using fuzzy models is developed. The most suitable techniques which can be used in each construction phase (experiment design, system identification, selection of regressors, structure and parameter adjusting, and validation) are treated in detail (Ch.~4). The most popular fuzzy controller design techniques (model-free, adaptive model-based, fuzzy gain scheduling) are reviewed and analyzed. It is proved that a fuzzy controller will guarantee the same or better performance than the one provided by a PID controller (Ch.~5). Unconstrained and constrained fuzzy predictive control strategies based on the combination of fuzzy modeling and model predictive control are proposed. The key point of the proposed strategies is the reduction of an original complex nonlinear program to a simple quadratic program by exploiting specific structure of fuzzy models (Ch.~6). A promising method to construct robust nonlinear predictive controllers based on the Takagi-Sugeno fuzzy model is introduced. The method converts the original nonlinear min-max program into a robust quadratic program, which can be reduced to an efficiently solvable second-order cone program (Ch.~7). The summary of contributions and future perspectives are gathered in Chapter~8. Five appendices support the already extensive results of the chapters by proofs, explanations and illustrative examples. The book (263 pages, 138 figures, 95 references) is of interest to researchers in the field of data mining, artificial intelligence, modeling, and control. Also, the realistic examples provide good material to graduate students and engineers.