Control of complex and uncertain systems. New types of feedback (Q1578315)

This book covers many aspects of control theory, but its main contribution lies in attempting to establish new fundamentals for feedback design. It presents in a monographic way results obtained by its authors in the last two decades, published principally in Russian, so not easily available to the general readership. An apparent curiosity of this book is that, except for few monographs that appeared before 1970, the authors make no reference to any publication in the control area from outside the former Warsaw Pact territory. The book consists of two parts. Part I, entitled Principles underlying the design of automatic control systems, includes two chapters and contains rudiments of linear control theory. Chapter 1 is devoted to the stabilization of uncertain plants by linear controllers subject to external disturbances. Special attention is paid to the high gain approach. Chapter 2 deals with the synthesis of nonlinear controllers, concentrating on the variable structure systems approach. Both chapters are very clearly written, illustrated with numerous figures and diagrams, so this material could be useful in an undergraduate course of control. Part II of the book, entitled New types of feedback, formulates a new approach to feedback design in uncertain systems. This new approach, called the generation principle, results from an observation that when dealing with control problems under deficit of information, one should rather attempt at designing rules of action than actions themselves. This means that instead of using a program control, we need to look for a rule of transformation of the error signal into control (i.e. a feedback operator), next for a rule of designing feedback operators (a meta-rule), next for a rule of designing meta-rules (a meta-meta-rule), etc. In such a way a hierarchy of feedbacks is designed, able to solve the control problem. To implement the generation principle the authors proceed in accordance with another principle, called binarity principle, asserting that coordinate signals and operator signals are equally right signals in control systems. In this context, a traditional feedback operating on coordinate signals is referred to as unary, while new types of feedback treating equivalently signals and (properly parameterized) operators belong to the class of binary feedbacks. Consequently, 4 types of dynamical objects and feedbacks have been identified: C -- coordinate objects, CO -- coordinate-operator objects, OC -- operator-coordinate objects, and O -- operator objects, as well as: CF -- coordinate feedback, COF -- coordinate-operator feedback, OCF -- operator-coordinate feedback, and OF -- operator feedback, depending on what signals (coordinate, operator) are processed. Foundations of the new approach to feedback design are presented in Chapter 3. Exemplary solutions of the stabilization problem for simple linear systems using CO-feedback implemented as relay feedback, P or PI feedback, and integro-relay feedback are given in Chapter 4. Chapter 5 of the book is devoted to some extensions of variable structure systems theory towards allowing for higher-order sliding modes, so producing a smoother control action. An improvement of stabilization obtained by CO-feedback is accomplished in Chapter 6 by employing O-feedback. Chapter 7 augments the solution to the stabilization problem with OC-feedbacks: integral-relay and imploying second degree sliding modes. In this way there appears a full 4 level hierarchy of feedbacks. Physical explanations of the behaviour of binary feedback systems are provided in Chapter 8. Chapter 9 reviews diverse standard algorithms of signal differentiation, and afterwards introduces new algorithms resulting from the binary feedback approach. The book ends with Chapter 10 dealing with optimal stabilization of linear plants. The usefulness of binary feedback methods for optimal stabilization is demonstrated by examples. The second part of this book is written in the same style as the first part, with special care for clarity and accessibility, therefore it could serve as supplementary material in a course of robust and adaptive control. Truly, this book converges at many points to concepts and ideas scattered in more or less explicit way in contemporary literature on adaptive control. However, it is a strong belief of the reviewer that it has gone far beyond known ideas, provided new results, and opened an intriguing perspective on the feedback control of uncertain systems.