Elsevier

Systems & Control Letters

Volume 75, January 2015, Pages 14-19
Systems & Control Letters

Stability of positive linear switched systems on ordered Banach spaces

https://doi.org/10.1016/j.sysconle.2014.10.004Get rights and content

Abstract

We provide sufficient criteria for the stability of positive linear switched systems on ordered Banach spaces. The switched systems can be generated by finitely many bounded operators in infinite-dimensional spaces with a general class of order-inducing cones. In the discrete-time case, we assume an appropriate interior point of the cone, whereas in the continuous-time case an appropriate interior point of the dual cone is sufficient for stability. This is an extension of the concept of linear Lyapunov functions for positive systems to the setting of infinite-dimensional partially ordered spaces. We illustrate our results with examples.

Introduction

Stability problems of switched systems have been studied in various works for the past years. Most authors focus on switched systems in Rd, see e.g.  [1], [2] and references therein. However, stability problems of switched systems in infinite-dimensional spaces have been addressed less frequently, see e.g.  [3]. Usually, there are two approaches when dealing with switched systems: Either analysing stability behaviour of the switched system under arbitrary switching or finding a switching signal which generates a stable solution, referred to as stabilizability of the switched system.

In this paper, we want to investigate the stability behaviour of positive linear switched systems. These systems have recently attracted a large interest due to the extensive possibilities of their application, e.g. in biology, economy or engineering (see e.g.  [4], [5] and the references therein). In the literature, stability of switched linear systems is proved by using different methods, e.g. the joint spectral radius  [6], [7], [8] or quadratic Lyapunov functions  [1]. Note that continuous-time switched systems can be equivalently formulated as differential inclusions, see e.g.  [9], [10], [11], [12], [13].

Considering positivity in Rd with respect to the standard order induced by the cone R+d{(x1,,xd)Rd;xi0  for  i=1,,d}, the existence of a linear Lyapunov function (LLF) is sufficient for the stability of a positive switched system, since the trajectories are restricted to the cone. Such an LLF can be constructed by means of an interior point of R+d which is mapped appropriately (see e.g.  [14]). In  [15], [16], the existence of an LLF is related to the product property of matrices. An algorithm to construct an LLF is given in  [17] using Collatz–Wielandt sets. Concerning positive linear switched systems with respect to polyhedral cones, we refer to  [18] for stability criteria using an LLF and to  [19] for a copositivity approach. For positive systems with respect to more general cones in Rd, stability is investigated in  [20], [21], where extremal norms are used.

In the present work we establish sufficient criteria for the stability of positive linear switched systems generated by bounded operators in infinite-dimensional spaces with a general class of order-inducing cones. In the discrete-time case, we exploit the idea of the existence of an appropriately mapped interior point of the cone to ensure stability of the switched system. A dual concept using a uniformly positive linear functional, i.e. an interior point of the dual cone, is established to prove stability in the continuous-time case. This is an extension of the concept of LLF cited above to infinite-dimensional partially ordered spaces. Even in the setting of finite-dimensional spaces ordered by arbitrary cones to our knowledge similar results have not been known so far.

The paper is structured as follows: In Section  2, we collect basic notions and concepts concerning ordered vector spaces and positive operators. Section  3 deals with positive linear discrete-time switched systems. We provide a criterion for the stability of such systems in order unit spaces and relate it to the joint spectral radius. In Section  4 we consider positive linear continuous-time switched systems on ordered Banach spaces. The stability of the system is ensured by the existence of a suitable uniformly positive linear functional. The main results are illustrated by examples.

Section snippets

Preliminaries

We recall basic notions concerning partial orders in vector spaces (see e.g.  [22], [23]).

Let X be a vector space over R. A set KX is called a wedge, if for every x,yK and for every λ,μ[0,) one has λx+μyK. If, in addition, for every xK with xK one has x=0, then K is called a cone. A cone K in X induces a partial order on X by xy:yxK.X with the order induced by K is called Archimedean, if x,yX:(nN:nxy)x0.X is called directed, if for every x,yX there is a zX such that zx

Positive linear discrete-time switched systems

In this section, let I{1,,N} be a fixed finite index set. Assume that (X,) is a normed space and A{A1,,AN}L(X) is a finite set. A map σ:N0I is called switching signal and the set of all switching signals is denoted by Σ. For x0X and σΣ consider the equation xk+1=Aσ(k)xkfor all  kN0.

Definition 7

Eq. (2) (together with Σ and A) is called linear discrete-time switched system. By SA{(xk)kN0X;σΣ  such that   (2)   holds} we denote the set of solutions to   (2), i.e. the set of all sequences (xk)k

Positive linear continuous-time switched systems

In this section, let I{1,,N} be a fixed finite index set. The map σ:[0,)I is called a switching signal, if the following properties hold:

  • (i)

    σ has only finitely many jumps (switching instances) on every bounded interval.

  • (ii)

    σ is left-continuous.

The set of all switching signals is denoted by Σ.

By making use of this definition, we are now able to define a continuous-time switched system by adapting  [3, Eqs. (9), (10)] (given for strongly continuous operator semigroups) for our case of bounded

Acknowledgements

The authors would like to thank the referees for their constructive comments which led to an improvement of the paper. The work of the first author is funded by Vietnam National Foundation for Science and Technology Development (NAFOSTED) under grant number FWO.101.2013.02. The work is partly supported by the German Research Foundation (DFG) through the Cluster of Excellence (EXC 1056), Center for Advancing Electronics Dresden (cfaed).

References (37)

  • D. Liberzon
  • H. Lin et al.

    Stability and stabilizability of switched linear systems: A survey of recent results

    IEEE Trans. Automat. Control

    (2009)
  • F.M. Hante et al.

    Converse Lyapunov theorems for switched systems in Banach and Hilbert spaces

    SIAM J. Control Optim.

    (2011)
  • L. Farina et al.

    Positive Kubear Systems. Theory and Applications

    (2000)
  • L. Fainshil et al.

    A maximum principle for the stability analysis of positive bilinear control systems with applications to positive linear switched systems

    SIAM J. Control Optim.

    (2012)
  • R. Jungers
  • V. Blondel et al.

    Polynomial-time computation of the joint spectral radius for some sets of nonnegative matrices

    SIAM J. Matrix Anal. Appl.

    (2009)
  • D. Angeli et al.

    Uniform global asymptotic stability of differential inclusions

    J. Dyn. Control Syst.

    (2004)
  • Cited by (4)

    • Stabilization of positive switched delay systems with all modes unstable

      2018, Nonlinear Analysis: Hybrid Systems
      Citation Excerpt :

      In particular, positive switched systems [3,4], which consist of a group of positive systems and a switching signal specifying the switching rules, have attracted considerable interests in the field of control. Evidence to date indicates that more and more experts and scholars have began to study the stability problem of positive switched systems, see [5–12]. The methods of common co-positive Lyapunov function [3,4], multiple co-positive Lyapunov functions [7], co-positive polynomial Lyapunov function [11] and joint linear co-positive Lyapunov function [12] are effectively adopted to investigate the stability problems of positive switched systems.

    • Optimal control of discrete-time bilinear systems with applications to switched linear stochastic systems

      2016, Systems and Control Letters
      Citation Excerpt :

      A common problem for a switched linear system is that of determining whether it is stable under arbitrary switching. Much effort has been made to approach this problem, resulting in various methods and tools; see the recent survey paper [9,10] and the references therein. This problem is closely related to determine the joint spectral radius (JSR) of the set of subsystems.

    View full text