SearcharxivSearch

arXiv subjects

Timothy H. Hughes

Publications and source records attributed to Timothy H. Hughes.

7 recordsLinked to original sources

Dissipativity, reciprocity and passive network synthesis: from Jan Willems' seminal Dissipative Dynamical Systems papers to the present day

The dissipativity concept sits at the intersection of physics, systems theory, and control engineering, as a natural generalisation of passive systems that dissipate energy. It relates the external behavior of systems to their internal state, and connects the subjects of optimal control, algebraic Riccati equations, linear matrix inequalities, complex functions, and spectral factorization. Within control, its applications include the analysis and design of interconnected systems (such as cyber-physical systems), robustness, and the absolute stability problem, and network synthesis (of electrical, mechanical, and multi-physics systems). This article details recent developments in the treatment of dissipativity and the related concept of reciprocity for systems that are not necessarily controllable and need not lend themselves naturally to an input-state-output perspective, as is the case for many physical and passive systems. We illustrate these concepts using simple electric circuit and mechanical network examples.

math.OC

Minimal series-parallel network realizations of bicubic impedances

An important open problem in the synthesis of passive controllers is to obtain a passive network that realizes an arbitrary given impedance function and contains the least possible number of elements. This problem has its origins in electric circuit theory, and is directly applicable to the cost-effective design of mechanical systems containing the inerter. Despite a rich history, the problem can only be considered solved for networks that contain at most two energy storage elements, and in a small number of other special cases. In this paper, we solve the minimal network realization problem for the class of impedances realized by series-parallel networks containing at most three energy storage elements. To accomplish this, we develop a novel continuity-based approach to eliminate redundant elements from a network.

eess.SY

On reciprocal systems and controllability

In this paper, we extend classical results on (i) signature symmetric realizations, and (ii) signature symmetric and passive realizations, to systems which need not be controllable. These results are motivated in part by the existence of important electrical networks, such as the famous Bott-Duffin networks, which possess signature symmetric and passive realizations that are uncontrollable. In this regard, we provide necessary and sufficient algebraic conditions for a behavior to be realized as the driving-point behavior of an electrical network comprising resistors, inductors, capacitors and transformers.

eess.SY

On the internal signature and minimal electric network realizations of reciprocal behaviors

In a recent paper, it was shown that (i) any reciprocal system with a proper transfer function possesses a signature-symmetric realization in which each state has either even or odd parity; and (ii) any reciprocal and passive behavior can be realized as the driving-point behavior of an electric network comprising resistors, inductors, capacitors and transformers. These results extended classical results to include uncontrollable systems. In this paper, we establish new lower bounds on the number of states with even parity (capacitors) and odd parity (inductors) for reciprocal systems that need not be controllable.

eess.SY

On the optimal control of non-expansive or passive systems

The positive-real and bounded-real lemmas solve two important linear-quadratic optimal control problems for passive and non-expansive systems, respectively. The lemmas assume controllability, yet a passive or non-expansive system can be uncontrollable. In this paper, we solve these optimal control problems without making any assumptions. In particular, we show how to extract the greatest possible amount of energy from a passive but not necessarily controllable system (e.g., a passive electric circuit) using state feedback. A complete characterisation of the set of solutions to the linear matrix inequalities in the positive-real and bounded-real lemmas is also obtained. In addition, we obtain necessary and sufficient conditions for a system to be non-expansive that augment the bounded-real condition with new conditions relevant to uncontrollable systems.

eess.SY

A theory of passive linear systems with no assumptions

We present two linked theorems on passivity: the passive behavior theorem, parts 1 and 2. Part 1 provides necessary and sufficient conditions for a general linear system, described by a set of high order differential equations, to be passive. Part 2 extends the positive-real lemma to include uncontrollable and unobservable state-space systems.

eess.SY

Why RLC realizations of certain impedances need many more energy storage elements than expected

It is a significant and longstanding puzzle that the resistor, inductor, capacitor (RLC) networks obtained by the established RLC realization procedures appear highly non-minimal from the perspective of linear systems theory. Specifically, each of these networks contains significantly more energy storage elements than the McMillan degree of its impedance, and possesses a non-minimal state-space representation whose states correspond to the inductor currents and capacitor voltages. Despite this apparent non-minimality, there have been no improved algorithms since the 1950s, with the concurrent discovery by Reza, Pantell, Fialkow and Gerst of a class of networks (the RPFG networks), which are a slight simplification of the Bott-Duffin networks. Each RPFG network contains more than twice as many energy storage elements as the McMillan degree of its impedance, yet it has never been established if all of these energy storage elements are necessary. In this paper, we present some newly discovered alternatives to the RPFG networks. We then prove that the RPFG networks, and these newly discovered networks, contain the least possible number of energy storage elements for realizing certain positive-real functions. In other words, all RLC networks which realize certain impedances contain more than twice the expected number (McMillan degree) of energy storage elements.

eess.SY