System identification and adaptive control. Theory and applications of the neurofuzzy and fuzzy cognitive network models (Q346644)

In this book the theory of neuro-fuzzy and fuzzy cognitive network models is presented together with their applications. It consists of two parts. In the first part, recurrent neuro-fuzzy models are studied. Different classes of adaptive control schemes are considered which are obtained by the choice of the parameter estimator, the choice of the control law, and the way they are combined. The adaptive control problem for nonlinear systems is studied. A new method, called parameter hopping, is introduced in order to present a novel neuro-fuzzy adaptive controller. This method allows the stability and control performance of the closed-loop system to be guaranteed. The neuro-fuzzy approximation model is constructed in order to control the systems. To do this the authors use Recurrent High-Order Neural Networks (RHONNs) together with approximately defined fuzzy dynamical systems. Introduction to RHONNs and the adaptive fuzzy systems is provided focusing mainly in the Mamdani centroid of area defuzzification procedure. At the end (Chapter 6) of Part I selected applications are presented such as trajectory tracking of robot manipulators, bioreactor application, experimental setup and neuro-fuzzy regulation of a DC motor. Many simulation results are presented as well. Part II of the book deals with Fuzzy Cognitive Networks (FCNs). Basic definitions and operations are given, as well as specific peculiarities appearing in their convergence. Convergence conditions and weight updating laws are derived. Existence of equilibrium points of FCNs which are equipped with continuous differentiable sigmoid functions is proved. Moreover, conditions on the weight interconnections related to the size of the FCN and the inclination of the sigmoid functions are obtained. These conditions allow the derivation of two adaptive estimation algorithms of the FCN parameters that fulfilled the FCN equilibrium equation. Simulations are presented which demonstrate the effectiveness of these algorithms. Several applications as a conventional benchmark plant and two real life potential projects to energy production from renewable power sources are provided.