- Research
- Open access
- Published:
Stability analysis of linear positive systems with time delays on time scales
Advances in Difference Equations volume 2012, Article number: 56 (2012)
Abstract
This article studies the necessary and sufficient conditions for asymptotic stability of the linear positive systems with bounded time-varying delays on time scales. Using Lyapunov functions, asymptotic stability conditions for positive systems with constant delays on time scales are established. The relations between the two types of systems have been found out, which bridged the gap between their stability. These results not only unify the existing ones of linear positive continuous- and discrete- systems but also contain more complicated time scales.
1 Introduction
In the real world, many physical, biological and medical systems involve variables that have nonnegative sign, e.g., population levels, absolute temperature, and concentration of substances. Such systems are referred to as positive systems. A dynamical system is called positive if any trajectory of the system starting from non-negative initial states remains forever non-negative for non-negative controls. An overview of state of the art in positive systems theory is given in the monographs [1–3].
Stability is one of the most important properties of dynamic systems, and a massive monograph has been concentrated on this topic for positive systems [4–9]. In the last few years, conditions for stability and robust stability of linear positive discrete-time systems with delays have been given in [10–19].
Firstly, we introduce some notations. denotes the set of positive (nonnegative) integers, denotes the set of real (nonnegative) numbers, is n-dimensional linear vector space over real (nonnegative, positive) numbers with the maximum modulus norm ∥·∥ given by ∥x∥ = maxi=1,2,···,n|x i |, , and is the set of all real (nonnegative) matrices of (n × m)-dimension. For , we write to indicate that all elements of matrix A are nonnegative (non-positive, positive, and negative) and the notation of matrices will be used throughout this brief: A = [a lg ] and
In [9], Haddad and Chellaboina obtained a necessary and sufficient stability condition for the following continuous-time positive systems with constant delays:
where for l ≠ g, τ ≥ 0, and is the vector value initial function. They obtained the following result [9, Theorem 3.1]: the above continuous-time positive system is asymptotically stable if and only if there exists a vector such that
Recently, Liu et al. [19] employed a comparison result to establish some necessary and sufficient conditions for the following discrete-time positive systems with time-varying delays:
where for i ∈ {0, 1, 2,..., p}, 0 ≤ τ i (t) ≤ τ i with , τ = max{τ i |i = 1, 2,..., p}, and is the vector-valued initial function. They found out the following similar result [19, Theorem 1]: the above discrete-time positive system is asymptotically stable if and only if there exists a vector such that
Since time scales is a tool for establishing a unified framework for continuous and discrete analysis. In the present article, we try to unify the necessary and sufficient conditions for asymptotic stability of linear positive continuous- and discrete-time systems with delays into one form on time scales.
This article is organized as follows. In Section 2, necessary preliminaries are presented, and some lemmas are provided. Section 3 proposes the necessary and sufficient stability criteria for positive systems with time-varying delays on time scales.
2 Notations and preliminaries
In this section, some basic definitions and some fundamental results are introduced, which are necessary for developing the main results of this article. Let denote a time scales, that is, is a nonempty closed subset of Denote and . For simplicity, let , , and , where n and p are positive integers.
Define the forward and backward jump operators by
where A point is called right-scattered, right-dense, left-scattered, and left-dense if σ(t) > t, σ(t) = t, ρ(t) < t, and ρ(t) = t, respectively. We put if is unbounded above and otherwise. The graininess is defined by
Let f be a function defined on f is said to be (delta) differentiable at provided there exists a constant a such that for any ε > 0, there is a neighborhood U of t (i.e., for some δ > 0) with
In this case, denote fΔ(t) := a. If f is (delta) differentiable for every , then f is said to be (delta) differentiable on If f is differentiable at , then
If FΔ(t) = f(t) for all , then F(t) is called an antiderivative of f on . In this case, define the delta integral by
Moreover, a function f defined on is said to be rd-continuous if it is continuous at every right-dense point in and its left-sided limit exists at every left-dense point in .
Definition 1. A function is said to be right-increasing at provided
-
(i)
f(σ(t0)) > f(t0) in the case that t0 is right-scattered;
-
(ii)
there is a neighborhood U of t0 such that f(t) > f(t0) for all t ∈ U with t > t0 in the case that t0 is right-dense.
If the inequalities for f are reverse in (i) and (ii), f is said to be right-decreasing at t0.
The following result can be directly derived from (1).
Lemma 1. Assume that is differentiable at . If fΔ(t0) > 0, then f is right-increasing at t0; and if fΔ(t0) < 0, then f is right-decreasing at t0.
Consider the following delayed system
where for , , , for , , and is the vector value initial function.
Definition 2. System (2) is said to be positive if and only if for any initial condition , the corresponding trajectory holds for all .
Lemma 2. If for all and for all then system (2) is positive.
Lemma 3. If system (2) is positive then for all and for l≠ g.
Represent (2) componentwise, i.e.,
where x l (·) is the l th component of vector x(·).
Proof of Lemma 2. Assume that for all , for all , and the initial condition . Then, we have by (3) and therefore, x l (t) ≥ 0 for .
Assume for contradiction that there exists such that x l (t1) < 0. Thus, there exists such that x j (t0) ≥ 0, x j (t0 - τ i (t0)) ≥ 0 for all , , and . It follows from (3) that
which contradicts to . Therefore system (2) is positive. This completes the proof.
Proof of Lemma 3. Suppose that system (2) is positive. By Definition 2, for all if .
Case 1. Assume for contradiction that there exists such that does not hold. Thus, there exists an element . Denote φ(-τ i (0)) = [φi 1, φi 2,..., φ in ]Tand x(0) = [φ01, φ02,..., φ0n]T. Choose such that φ qg > 0 and φ ij = 0 for all i ≠ q or j ≠ g. Since q ≠ 0 then x(0) = [φ01, φ02,..., φ0n]T = [0, 0,..., 0]T. It follows from (3) that
Since x l (0) = 0 and , then there exists such that x l (t0) < 0, which contradicts the assumption that for all .
Case 2. Assume for contradiction that for l ≠ g. Let φ(-τ i (0)) = [0, 0,..., 0]T and x(0) = [0,..., 0, φ0g, 0,..., 0]T where φ0g> 0. It follows from (3) that
Since x l (0) = 0 and , then there exists such that x l (t0) < 0, which contradicts the assumption that for all . Therefore, for all and for l ≠ g. This completes the proof.
Lemma 4. Assume that system (2) is positive. Let x a (t) and x b (t) be the trajectories solutions of (2) under the initial conditions φ a (·) and φ b (·), respectively. Then implies that for all .
Proof. It is easy to see that x(t) = x b (t) - x a (t) is the solution to system (2) under the initial condition . Since (2) is positive, for all . This completes the proof.
3 Main results
The purpose of this section is to establish asymptotic stability criteria for the positive system with bounded time-varying delays on time scales. Consider the following positive system
where for , for , for , and the delays satisfy
with constants , , and
is the vector-valued initial function. By Lemma 2, system (4) is positive.
Now, we consider the following system closely related to (4)
where all the system matrices are as in (4), and τ i in (7) is the supremum of τ i (t) in (4), as shown in (5), τ is defined by (6), and is an arbitrary initial condition. Note that the delays in (7) are constant.
Lemma 5. Consider system (7). Suppose that there exists a vector satisfying
and that the initial condition is ϕ(·) ≡ λ. Then, for any , the solution to system (7) satisfies
Proof. Note that the initial condition ϕ(·) ≡ λ. By (7), it yields that
It follows from y(t) = ϕ(t) ≡ λ for all , (8), and (10) that for any .
Now assume for contradiction that there exists such that yΔ(t0) ϕ 0 and for any . By Lemma 1, y(t) is right-decreasing for all .
It follows from (7), (8), and y(t) = ϕ(t) ≡ λ for that
which contradicts the assumption that yΔ(t0) ϕ 0. Therefore, for any This completes the proof.
The next lemma will show that, under certain conditions, the solution to system (4) is not greater than that to system (7), as stated in the following discussion.
Lemma 6. Suppose that there exists a vector satisfying (8) and the initial conditions for system (4) and (7) are the same, i.e., φ(·) ≡ ϕ(·) ≡ λ. Then
for all , where x(t) and y(t) are solutions to (4) and (7), respectively.
Proof. By (4) and (7), it follows that
and
Hence xΔ(0) = yΔ(0), which implies x(σ(0)) = y(σ(0)).
Moreover, there exists such that x(t) = y(t) for all . By Lemma 5 and (5), we have , , and
which implies . In the following, we can prove that for all .
Assume for contradiction that there exists such that x(t2) ϕ y(t2). Thus, there exists such that for all and xΔ(t1) ϕ yΔ(t1). It follows from (4), (5), (7), and Lemma 5 that
which contradicts the assumption xΔ(t1) - yΔ(t1) ϕ 0. Then (11) holds for all . This completes the proof.
Consider the system (7) satisfying the assumption y(t) = ϕ(t) ≢0 for all .
Lemma 7. Let (7) be a positive system. Then the positive system (7) is asymptotically stable if and only if there exists a vector such that (8) holds.
Proof. Necessity: By integration, it follows from (7) that
Then
Since the system (7) is asymptotically stable, then y(t) → 0 for t → +∞. It follows from
that
From assumption that y(t) ≢ 0 for all and (7) is a positive system, we have the left side of (12) is negative and
The condition (13) is equivalent to for
Sufficiency: Let us consider the dual system
which is positive and asymptotically stable.
As a Lyapunov function for the dual system (14) we may choose the following function
which is positive for non-zero and for positive .
From (14) and (15) we have
Hence, the condition implies {V[x(t)]}Δ ≺ 0 and the positive system (7) is asymptotically stable. This completes the proof.
Theorem 1. Let (4) be a positive system. Then the positive system (4) is asymptotically stable if and only if there exists a vector such that (8) holds.
Proof. Sufficiency: Assume that there exists a vector such that (8) holds. By Lemma 7, system (7) is asymptotically stable.
Define . Since system (7) is asymptotically stable, for any given ε > 0, there exists a scalar δ > 0, such that if ∥ϕ∥ < δ, then the corresponding solution y ϕ (t) to (7) satisfies that ∥y ϕ (t)∥ < ε for all and limt→+∞∥y ϕ (t)∥ = 0.
In particular, choose the initial condition ϕ(·) = αλ, where α > 0, such that ∥αλ∥< δ. Then, the corresponding solution y αλ (t) must satisfy that ∥y αλ (t)∥ < ε for all and limt→+∞∥y αλ (t)∥ = 0.
For system (4), let the initial condition be φ(·) = αλ. By Lemma 6, the corresponding solution x αλ (t) satisfies for all . In addition, by Lemma 4, for any initial condition , the corresponding solution satisfies for all .
Therefore, for system (4), arbitrary given ε > 0, there exists δ1 = ∥αλ∥ > 0 such that ∥φ∥ < δ1, implies ∥x φ (t)∥ ≤ ∥x αλ (t)∥ ≤ ∥y αλ (t)∥ < ε for all .
Furthermore, since ∥x φ (t)∥ ≤ ∥ y αλ (t)∥ for all , limt→+∞∥y αλ (t)∥ = 0 implies
and therefore, system (4) is asymptotically stable.
Necessity: Suppose that system (4) is asymptotically stable for any delays satisfying (5). Particularly, let τ i (t) = τ i , then system (7) is necessarily asymptotically stable. It follows from Lemma 7 that (8) must hold. This completes the proof.
For the continuous-time positive systems with constant delays and discrete-time positive systems with time-varying delays, [9, 19] have established some necessary and sufficient stability criteria. Theorem 1 can be viewed as an extension to time scales.
Remark 1. In the continuous case: µ(t) ≡ 0 (i.e., ). It follows from (1) that the positive system (4) can be written as:
where 0 ≤ τ i (t) ≤ τ i , τ = max{τ i |i = 1, 2,..., p}, and is the vector-valued initial function. Let A0 - I = A, p = 1, A1 = A d , and τ i (t) ≡ τ, then the system (16) is the same as the continuous-time positive system with constant delays in [9]. Hence, Theorem 1 is the same as that obtained by Haddad and Chellaboina [9, Theorem 3.1] (see Section 1).
Remark 2. In the discrete case: µ(t) ≡ 1 (i.e., ). It also follows from (1) that the positive system (4) can be written as:
where 0 ≤ τ i (t) ≤ τ i , τ = max{τ i |i = 1, 2,..., p}, and is the vector-valued initial function. By Δx(t) = x(t + 1) - x(t), the above system can be rewritten as:
It is evident that system (17) is of the same form as the discrete-time positive system with time-varying delays in [19]. Hence, Theorem 1 in this special case is the same as [19, Theorem 1] (see Section 1).
Remark 3. Since system (4) includes delay-independent case (i.e., τ i (t) ≡ 0), if positive system (4) is asymptotically stable, then the matrix in (8) is a Hurwitz matrix.
Example 1. The positive systems play a key role in understanding many processes in biological science. To illustrate this, a delay-independent model of glycolysis in yeast has been constructed. The glycolysis is the name of a cellular process in which glucose is converted into different carbon compounds in a number of reaction steps. This process plays an important role in the metabolism of most organisms, and its function is partly to produce energy rich molecules and partly to produce carbon compounds that can be used for biosynthesis. The glycolysis has been extensively explored in many organisms and several mathematical models of it have been made. In 2000, a model of glycolysis has been made by Teusink et al. [20]. In order to make this model suitable for our system, a few modifications have been done. Such as changing state variables to constants and changing certain constants into state variables. After these changes the model contains 13 state variables, x1,..., x13, each representing the concentration of a carbon compound, except from one, x5, which represents a module of two interconvertable carbon compounds. Since the biological process is nonlinear, the system was linearized around its steady state.
where x = (x1,..., x13)T is the vector of the 13 state variables and
Since A is a Hurwitz matrix and all its eigenvalues lie in the left-half of the complex plane, see Table 1, the positive system (18) is asymptotically stable. Let
then we have
By Theorem 1, we can also deduce that (18) is an asymptotically stable positive system.
References
Berman A, Neumann M, Stern RJ: Nonnegative Matrices in Dynamic Systems. New York, Wiley; 1989.
Kaczorek T: Positive 1D and 2D Systems. Springer-Verlag, London; 2002.
Kaczorek T: Polynomial and Rational Matrices, Applications in Dynamical Systems Theory. Springer-Verlag, London; 2007.
Gorecki H: Analysis and Synthesis of Control Systems with Delay. WNT, Warszawa; 1971.
Gorecki H, Fuksa S, Grabowski P, Korytowski A: Analysis and Synthesis of Time Delay Systems. PWN-J. Wiley, Warszawa-Chichester; 1989.
Gorecki H, Korytowski A: Advances in Optimization and Stability Analysis of Dynamical Systems. Publishing Department of University of Mining and Metallurgy, Krakow; 1993.
Buslowicz M: Robust Stability of Dynamical Linear Stationary Systems with Delays. Publishing Department of Technical University of Bialystok, Warszawa-Bialystok; 2000.
Niculescu SI: Delay E ects on Stability. A Robust Control Approach. Springer-Verlag, London; 2001.
Haddad WM, Chellaboina V: Stability theory for nonnegative and compartmental dynamical systems with time delay. Syst Control Lett 2004, 51: 355–361. 10.1016/j.sysconle.2003.09.006
Hinrichsen D, Hgoc PHA, Son NK: Stability radii of positive higher order difference systems. Syst Control Lett 2003, 49: 377–388. 10.1016/S0167-6911(03)00116-6
Buslowicz M, Kaczorek T: Stability and robust stability of positive linear discrete-time systems with pure delay. In Proc 10th IEEE Int Conf on Methods and Models in Automation and Robotics, Miedzyzdroje. Volume 1. IEEE, Piscataway; 2004:105–108.
Buslowicz M: Stability of positive singular discrete-time systems with unit delay with canonical forms of state matrices. In Proc 12th IEEE Int Conf on Methods and Models in Automation and Robotics, Miedzyzdroje. Volume 2006. IEEE, Piscataway; 2006:215–218.
Buslowicz M: Robust stability of positive discrete-time linear systems with multiple delays with linear unity rank uncertainty structure or nonnegative perturbation matrices. Bull Pol Acad Sci Tech Sci 2007, 55: 1–5.
Rami MA, Tadeo F, Benzaouia A: Control of constrained positive discrete systems. In Proc Am Control Conf, New York. Volume 2007. IEEE, Piscataway; 2007:5851–5856.
Rami MA, Tadeo F: Positive observation problem for linear discrete positive systems. In Proc 45th IEEE Conf Decision Control, San Diego. Volume 2006. IEEE, Piscataway; 2006:4729–4733.
Hmammed A, Benzaouia A, Rami MA, Tadeo F: Positive stabilization of discrete-time systems with unknown delay and bounded controls. In Proc Eur Control Conf, Kos. Volume 2007. European Union Control Association, Zurich; 2007:5616–5622.
Buslowicz M: Simple stability conditions for linear positive discrete-time systems with delays. Bull Pol Acad Sci Tech Sci 2008, 56: 325–328.
Yu W, Zhong S: Constrained control of positive discrete-time systems with delays. IEEE Trans Circ Syst II Exp Briefs 2008, 55: 193–197.
Liu X, Yu W, Wang L: Stability analysis of positive systems with bounded time-varying delays. IEEE Trans Circ Syst II Exp Briefs 2009, 56: 600–604.
Teusink B, Passerge J, Reijenga CA, Esgalhado E, van der Weijden CC, Schepper M, Walsh MC, Bakker BM, van Dam K, Westerhoff HV, Snoep JL: Can yeast glycolysis be understood in terms of in vitro kinetics of the constituent enzymes? testing biochemistry. Eur J Biochem 2000, 267: 5313–5329. 10.1046/j.1432-1327.2000.01527.x
Acknowledgements
Many thanks to Leonid Berezansky (the editor) and the anonymous reviewers for helpful comments and suggestions. This research was supported by the NNSF of China (Grant Nos. 11071143, 11101241, and 61174217), the NNSF of Shandong Province (Grant Nos. ZR2009AL003, ZR2010AL002, ZR2010AL016, ZR2011AL007, and JQ201119), the Scientific Research and Development Project of Shandong Provincial Education Department (J11LA01), and the NSF of University of Jinan (XKY0918).
Author information
Authors and Affiliations
Corresponding author
Additional information
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
YS directed the study and helped inspection. CZ carried out the the main results of this article, established the model, and drafted the manuscript. All the authors read and approved the final manuscript.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
About this article
Cite this article
Zhang, C., Sun, Y. Stability analysis of linear positive systems with time delays on time scales. Adv Differ Equ 2012, 56 (2012). https://doi.org/10.1186/1687-1847-2012-56
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/1687-1847-2012-56