Loop transfer recovery. Analysis and design (Q1917978)

The book deals with several issues of analysis and design of what is known as a loop transfer recovery (LTR) problem. LTR is developed as an attempt to improve the robustness of observer based controllers. Controllers utilizing complete state feedback designed via linear quadratic optimal control theory have several advantages. However, these advantages are generally lacking when a full state feedback controller is implemented via an observer based measurement feedback controller, unless the observer is designed properly. The usual guideline in designing an observer is to make its dynamics much faster than the desired state feedback dynamics so that the state reconstruction can be expedited. Such speeding up of observer dynamics, however, as shown by Doyle and Stain, does not in general attain the desired results. Since the works of Doyle and Stain, and Kwakernaak the problem of LTR has been investigated on several fronts. In the book, general non-minimum phase and non-strictly proper systems of both continuous as well as discrete type are considered. In the case of continuous-time systems LTR in general is not feasible if the given system is not left invertible and/or not minimum phase; in the case of discrete time-time systems, LTR in general is possible only for a restricted class of systems. One of the aims of the book is to analyze systematically the following issues concerning the LTR mechanism for general systems: (a) characterizing the available freedom in designing controllers for a given system and for an arbitrary given target loop transfer function when its specific characteristics are not taken into account, (b) development of necessary and/or sufficient conditions a target loop transfer function has to satisfy so that it can either exactly or asymptotically be recovered for a given system, (c) development of necessary and/or sufficient conditions on a given system such that it has at least one recoverable transfer function and (d) development of methods to test whether the intended recovery is possible in at least a given subspace of the control space or not. Another aim is to develop appropriate design tools for LTR or whatever recovery that can be achieved. In this regard, three methods of designing observer based controllers have been proposed. These methods are: an asymptotic eigenstructure and time scale assignment method, an \(H_2\) norm minimization method, and an \(H_\infty\) norm minimization method. The first four chapters of the book consider general continuous-time systems, and cover all the topics of LTR as it is conventionally defined. The next four chapters do the same except that they consider discrete-ti Chapter 9 formulates a new closed-loop transfer recovery problem. The last Chapter 10 introduces a new architecture, called a CSS architecture, for a controller. The appropriate background for this book is a first graduate course in linear systems and state-space methods with an elementary knowledge of linear quadratic control.