The mathematics of networks of linear systems (Q2339763)

This book is intended to provide readers with an algebraic systems theory perspective of the mathematical foundations of networks of linear systems. The approach is based on polynomial model representations of linear systems, given by coprime factorizations of the transfer function of the system. Over the years two different approaches have been developed to deal with large-scale interconnected systems: decentralized control, as described in the book of Siljak (1991), where a multivariable plant is to be controlled by a structured controller and the study of feedback connections, as, for instance, in the analysis of switching circuits (Leonard and Krishnaprasad (1994)). In practical implementations of network control systems, models for the node systems may be represented by first-order or higher-order difference and differential equations. First-order representations of both the node systems and the interconnection dynamics allow to apply standard controllability and observability tests for closed-loop systems, but this approach quickly becomes inefficient to deal with medium-scale networks of high dimensional subsystems. To investigate networks of linear systems, the authors propose an approach based on polynomial matrix fraction representations. A key tool is the principle of permanence of the strict system equivalence of networks, reformulated and extended to dynamic interconnections within the unifying framework of polynomial models. Techniques based on the shift realization, as in Fuhrmann (1976), allow to analyse controllability properties of higher-order systems and controllability information can be used to reduce the complexity of computations. The book includes a thorough treatment of controllability, observability and realization theory, as well as feedback control and observer theory. The material is quite extensive and is not intended as a textbook for a single course. Part I can be used for a first course in algebraic systems theory. The approach allows to deal in a unified way with state space, polynomial system matrices, geometric control methods and, suitably modified, it can be used in connection with the study of some infinite-dimensional systems. In Chapter 1, two central issues of the control of networks are presented: open-loop control of interconnected systems and synchronization by discussing some illustrative examples. In Chapter 2, there are collected some of the most relevant concepts and results from linear algebra that play a dominant role in the polynomial approach to linear systems. The reader is assumed to be familiar with the basic definitions and facts of linear algebra over a field. The theory of vector spaces and linear operators over a field is generalized to an analogous theory of modules, defined over the ring of polynomials in one variable with coefficients in a field. In Chapter 3, the authors show the advantage of replacing a linear transformation by an equivalent transformation that acts on a space of polynomials or rational functions, emphasizing also the computational aspects. Chapter 4 introduces basic definitions and concepts from linear systems theory, focusing on existence, uniqueness, and minimality of state-space realizations and on the notion of system equivalence. The approach, that will also be useful in later chapters to deal with interconnected systems, is developed from the point of view of functional models and shift operators. Part II could be used for a second, advanced, course on linear systems. Chapter 5 covers several topics that can be applied to the study of linear systems, such as bilinear forms defined on vector spaces, module homomorphisms over various rings, and the analysis of classes of structured matrices such as Bezoutian, Hankel, and Toeplitz matrices. The connections to algebraic methods for analyzing 2D systems such as modules over the ring of polynomials in two variables with coefficients in a field, are also explored. The unifying tools are tensor products defined for modules and vector spaces. Chapter 6 is devoted to analyze the effects of state feedback control and output injection on linear systems. The notion of controlled and conditioned invariant subspaces for a linear system, introduced and extensively studied in the early 1970s by Basile and Marro as well as Wonham and Morse in the context geometric control theory, are investigated. Chapter 7 is devoted to state estimation. The aim is to design an observer that uses information on observed variables and inputs in order to estimate linear functions of the state variables. Part III , which is largely independent of the previous parts, is ideally suited for advanced research seminars aimed at preparing graduate students for independent research. It addresses some of the basic research questions and approaches concerning networks of linear systems, such as, for example, synchronization, clustering, ensemble control and controlling infinite platoons. Chapter 8 is devoted to the description of connections and coupling patterns of dynamical systems in terms of graph theory and presents the main results and tools from matrix analysis and graph theory that will be used in subsequent chapters for analysing interconnected systems. In Chapter 9, the classical concept of strict system equivalence is used to prove that interconnection pattern between the node systems can be described by a weighted directed graph. From this principle are derived very concise and explicit characterizations of reachability and observability for homogeneous networks consisting of identical SISO systems. Further characterizations of reachability are obtained for special interconnection structures, such as paths, cycles, and circulant structures. Chapter 10 is devoted to several computational problems from Chapters 4 and 9 that are related to the concepts of reachability and observability. The focus is on the more common types of interconnection, such as parallel and series interconnections, and it is shown how these results are related to polynomial interpolation theory. Chapter 11 contains a fairly general approach to problems of synchronization and clustering for networks of identical linear systems. Simple autonomous models for mean value consensus are presented, then synchronization, for both first-order and higher-order system dynamics is considered. Chapter 12 provides an introduction to the emerging field of ensemble control for linear systems, namely the control of families of linear systems. Ensemble control refers to the task of controlling a large, potentially infinite, number of states, or systems, using a single-input function or a single-feedback controller. The focus is on the specific task of finding open-loop controls that steer a system from a family of initial states to another family of terminal states using a single open-loop control function that is independent of the parameters.