Control Systems with Input and Output Constraints

Abstract

It is difficult to overstate the importance of considering control constraints in control system design: such constraints have well-known implications for the behaviour of the resulting closedloop system, and ignoring these constraints can lead to a dramatic loss of performance and, potentially, stability. It is therefore not surprising that papers addressing these issues have populated the literature for many years, although the number of books largely devoted to constrained control systems is rather small. Notable exceptions are [1][2][3], although none of these works consider perhaps the most common way of addressing control constraints: anti-windup compensation.The subject of this review is the first and, so far, only, book to address the topic of anti-windup compensation, and it is authored by two eminent authorities in the field. Professors Glattfelder and Schaufelberger are regular contributors to the anti-windup literature and accomplished practical control engineers; it is this background which influences the style and tone of the book. In addition to input (control) constrained systems, the book treats systems having their outputs constrained in a similar manner. The approach adopted for the treatment of this topic is 'override' control, which is akin to anti-windup compensation for the constrained outputs rather than the constrained inputs. The authors are in the minority of academics who consider this important control engineering problem and it is the consideration of this topic which, perhaps, distinguishes the book above anything else.The book is essentially split into two parts: the first part covers PI control of dominant first order systems; the second part generalizes the concepts of the first to plants of higher order with more complex controllers and also lays the foundations for multivariable anti-windup/override control.An interesting feature of the book is the 'inductive' approach which the authors adopt; rather than starting from a general treatment of the problem, the authors prefer to point out the key features of their ideas using relatively simple systems at first and then extend these ideas to more complex systems. This strategy accounts for the first part of the book, where only first order systems under PI control are considered, and it works rather well. In the first part, the concepts of anti-windup, override control and simultaneous anti-windup and override control are introduced for systems with, respectively, input, output and input and output constraints. The emphasis in these chapters is on the architecture of the antiwindup/override controllers and a useful practical strand is developed as the authors describe and group various techniques which are currently in use in industry. The main goal of these chapters is to introduce the reader to the concepts associated with the design of such compensators and how the choice of the parameters associated with the antiwindup compensators/override controllers affects the transient response of the system.In the second part of the book,...

A non-square sector condition and its application in deferred-action anti-windup compensator design

“…Normally, the AW compensator is activated upon the occurrence of saturation. The subject of AW is now fairly mature and the reader is referred to the surveys [29,6], the edited volume [28] and the monographs [7,11,36]. In addition, two earlier papers [16,3] describe and connect some of the early work on AW.…”

Model predictive control of dynamically substructured systems with application to a servohydraulically actuated mechanical plant

“…[4], [6], [7] and the references therein). However, the problem of coping with actuator saturation within DSS has received scant attention.…”

A robust override scheme enforcing strict output constraints for a class of strictly proper systems

“…In this sense, the paper by Wolff and Buss (2004) considers a special class of non-linear single-input full-information feedback systems without strict consideration of disturbances. This severely contrasts the approaches by Glattfelder and Schaufelberger (1988), Glattfelder and Schaufelberger (2003) and in particular by Turner and Postlethwaite (2002). There it is easily possible to consider large scale multi-input-multi-output plants for output feedback control, disturbances and more than one output measurement constraint, while only 'soft output limits' are enforced.…”

“…Among those two issues, the problem of output signal constraints has been of lesser interest, although a significant amount of publications exists: Conceptual results have been presented by Saberi et al (2002) giving conditions on the existence of controllers for constrained systems. Other work aimed at controller design can be grouped into four areas: control using the 'maximum output admissible set' (Gilbert and Tan, 1991), model predictive control (Bemporad et al, 2003), invariance control (Mareczek et al, 2000;Mareczek et al, 2001;Mareczek et al, 2002;Wolff and Buss, 2004) and override control (Glattfelder and Schaufelberger, 1988;Turner and Postlethwaite, 2002;Glattfelder and Schaufelberger, 2003). The method for control using the 'maximum output admissible set' can be computationally very expensive as it is the aim to compute at first the maximum reachable set assuming certain plant input and output constraints.…”