Abstract
This paper deals with the stability of nonlinear fractional differential systems with delay. Based on the Lyapunov functional method and the Lyapunov function method, respectively, several stability criteria including Razumikhin-type stability criteria are derived, which are extensions of some existed results of the Hale and Verduyn Lunel (Introduction to functional differential equations. Springer, Berlin, 1993), Aguila-Camacho et al. (Commun Nonlinear Sci Numer Simul 19, 2951–2957, 2014) and Zhou et al. (Appl Math Lett 28, 25–29, 2014). In addition, some examples are provided to illustrate the applications of these criteria. Numerical simulations show the validity of our results.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Avoid common mistakes on your manuscript.
1 Introduction
In the past few decades, many researchers have devoted themselves to fractional calculus because many practical systems can be properly described with the help of fractional derivative. For the basic theory of fractional calculus and fractional differential systems (equations), one can see the monographs [1–6] and the references therein.
In the real dynamical systems, delayed phenomenon is unavoidable. Fractional delayed differential systems have received considerable attentions recently because they are more accurate in modeling memory phenomenon and hereditary phenomenon. In [7–14], the authors dealt with the existence of the solution of fractional functional differential systems (equations).
The stability of the equilibrium point of a system is of importance in control theorem. Stability criteria for linear fractional delayed differential systems (equations) can be found in [15–18]. Stability of nonlinear fractional delayed differential systems (equations) was investigated in [19–25].
Lyapunov’s direct method is an effective tool to analyze the stability of nonlinear integer-order differential systems without solving its state equations. It used to be thought that Lyapunov method is invalid to nonlinear fractional differential systems. However, this thought was proved to be wrong. In 2010, professor chen et al. proposed a Lyapunov function to deal with the stability of a class of nonlinear fractional differential systems [26]. In 2014, based on a lemma themselves and a theorem in the Ref. [26], Aguila-Camacho et al. [27] derived a simple stability criterion for a class of nonlinear fractional differential system. Corollary 1 in the Ref. [27] is similar to Theorem 3.1 in the Ref. [28].
Recently, Lyapunov function has been chosen to analyze the stability of nonlinear fractional differential systems with delay. For example, the authors in [20] proposed a Lyapunov function to deal with the Mittag–Leffler stability of the nonlinear fractional neutral differential system
where \(x(t)\in R^n\), \(0<q\le 1\), \(E, A\in R^{n\times n} \) are constant matrices and E may be singular \((rank E=n_1\le n )\). Further assume that matrix pair (E, A) is regular with index one, \(f:R\times (\hbox {bounded sets in C}) \) into bounded sets of \(R^n \). One of the results in [20] is as follows.
Suppose that the operator D is stable, and \(\alpha _1\), \(\alpha _2\) and \(\alpha _3\) are positive numbers. If there exists a continuous function \( V:[t_0, \infty )\times C\rightarrow R^n \) such that
-
(a)
\( \alpha _1\Vert x\Vert \le V(t,x) \le \alpha _2\Vert x\Vert \);
-
(b)
\( {\hbox {}^C}_{t_0}D_{t}^{\beta }V(t, Dx_t)\le -\alpha _3 \Vert Dx_t\Vert \), where \(0<\beta \le 1\), then the solution of system (1) is Mittag–Leffler stable.
By proposing a Lyapunov function and taking its integer derivative, the authors in [25] derived a stability criterion for the fractional delayed differential system
where \(x\in R^n\) is the state vector, \(f\in R^n \) is nonlinear vector function satisfying Lipschitz condition and the delay time \(\tau =[\tau _1,\tau _2, \ldots , \tau _n]\in R^n\) and \(t_0\in R^+\). The stability criterion of the Ref. [25] is as follows:
Suppose that fractional order \(0<\alpha \le 1\). If there is a positive definite matrix P and a semi-positive definite matrix Q, for any \(x(t)\in R^n\), fractional system (2) satisfies
the fractional delayed system (2) is Lyapunov stable.
Obviously, the stability criterion for the system (2) is inconvenient for application.
Lyapunov functional method is usually used to analyze the stability and designs of switching controller of nonlinear integer-order functional differential equations (systems) [29–32]. However, at present we have not retrieved any related references in which Lyapunov functional was used to investigate the stability of fractional differential systems with delay. Motivated by this fact, in this paper we mainly apply Lyapunov functional method to investigate the stability of nonlinear fractional delayed differential system of the form
with the initial state
where \(t_0\in R^+\), \({{\hbox {}^C}}D_{t_0,t}^{\alpha }\) is the Caputo’s fractional derivative of order \( \alpha \) with \(0<\alpha <1\), r is the delay satisfying \(r>0\), \(f:[t_0, +\infty )\times C([-r,0], R^n ) \rightarrow R^n\) is a given functional with \(f(t,0)=0\), \(\phi \in C([-r,0], R^n )\), \(f(t,\varphi )\) is continuous and satisfies Lipschitz condition in the second variable \(\varphi \) for \(\varphi \in C([-r,0], R^n )\). If \(x\in C([t_0-r, +\infty ), {R}^n)\), then for any \(t\in [t_0, +\infty )\) define \(x_t\) by \(x_t(\theta )=x(t+\theta )\), \(\theta \in [-r,0]\).
The simplest of the system (4) is fractional differential-difference equations of the type
where delay \(\tau >0\) is a constant. It is easy to see that Eq. (2) is a special case of Eq. (4). If A and E in Eq. (1) satisfy that \(E=I\)(unit matrix) and \(A=0\), then Eq. (1) is turned into Eq. (4). In other words, Eq. (4) is a special case of Eq. (1). However, the results of this article are different from ones in [20]. Furthermore, the main method used in this article is Lyapunov functional not the Laplace transform used in [20].
A precondition to study the stability of the equilibrium point \(x=0\) of the system (4) is that the system (4)–(5) must has a unique global solution defined on \([t_0-r, \infty )\). In one-dimensional space, the existence and uniqueness of the state solution defined on \( [t_0-r,\infty )\) of the system (4)–(5) can be guaranteed by Corollary 5.1 of the Ref. [13]. For the case \(R^n\) (\(n\ge 2\)), there is not any related result to guarantee that the system (4)–(5) has a unique solution defined on \([t_0-r, \infty )\). Throughout this paper, we always assume that the system (4)–(5) has a unique solution defined on \([t_0-r,\infty )\).
The contribution of this paper is that by choosing Lyapunov functionals and Lyapunov functions, respectively, we derived three stability criteria for the system (4)–(5) which are extensions of ones of the Ref. [29]. In addition, Corollary 1 of this article is simple and convenient for application, which is an extension of the results of the Refs. [27, 28].
The rest of this paper is organized as follows. Section 2 is devoted to some preliminaries. In Sect. 3, we present our main results of this article. Some illustrative examples and numerical simulations form the contents of Sect. 4.
2 Preliminaries
In this section, we recall some definitions and lemmas which will be used later.
Throughout this paper, notation \({R}^n\) denotes the n-dimensional Euclidean space, \(|\cdot |\) denotes the Euclidean vector norm, \((\cdot )^\mathrm{T} \) denotes the transpose of the vector \((\cdot )\), \( {\mathcal {K}}\) denotes the set of class-K functions. Denote \({\mathcal {C}}=C([-r,0], {R}^n )\) the space of continuous functions on \([-r,0]\). For any element \(\phi \in {\mathcal {C}}\), define the norm \(\Vert \phi \Vert =\max \limits _{-r\le \theta \le 0}|\phi (\theta )|_{{R}^n }\). For more details, readers may refer to the monograph [29].
Definition 1
A continuous function \(u(t):[0, +\infty )\rightarrow [0, +\infty )\) is said to be a class-K function if the function u(t) is strictly increasing and \(u(0)=0\).
Definition 2
[1] Given an interval [a, b] of R. The fractional order integral of a function \(f\in L^{1}([a,b], {R})\) of order \(\alpha \in {R}^+\) is defined by
where \(\varGamma \) is the Gamma function.
Property 1
Let \(0<\alpha <1\) and \(f(t)\ge 0\) on [a, b]. Then it holds
Definition 3
[1] Suppose that a function f is defined on [a, b] and \(f^{(n)}(t)\in L^1[a, b]\). The Caputo’s derivative of order \(\alpha \) with the lower limit a for the function f is defined as
where \(0<n-1<\alpha \le n \).
Particularly, when \(0<\alpha \le 1\), it holds
Lemma 1
[27] Let \(x(t)\in {R}^n\) be a continuous and derivable function. Then, for any instant \( t_0\in {R}\), it holds
Lemma 2
(Theorem 5.1 in [13]) Let \(\alpha \in {R}^{+}\), \(m=[\alpha ]+1\), and \(y(t)\in AC^m[0, b]\). Define the set \(G:=[0, +\infty )\times {\mathcal {C}}\) and let the function \(f:G\rightarrow {R}\) be continuous and fulfill a Lipschitz condition with respect to the second variable with a Lipschitz constant \(L\in {R}^+\) that is independent of t, \(y_1\) and \(y_2\). Then there exists a uniquely defined function \(y\in C[-r, b)\) solving the initial value problem
with initial conditions
where \([\alpha ]\) denotes the integer part of \(\alpha \).
Lemma 3
(Corollary 5.1 in [13]) Let \(\alpha \in {R}^{+}\), \(m=[\alpha ]+1\), and \(y(t)\in AC^m[0, b]\). Define the set \(G:=[0, +\infty )\times {\mathcal {C}}\) and let the function \(f:G\rightarrow {R}\) be continuous and fulfill a Lipschitz condition with respect to the second variable with a Lipschitz constant \(L\in {R}^+\). Then there exists a uniquely defined function \(y\in C[-r, \infty )\) solving the initial value problem (12)–(13).
3 Main results
In this section, three stability criteria which are parallel to ones of the Ref. [29] are derived by applying Lyapunov functional method and Lyapunov function method, respectively.
Theorem 1
Suppose that \(f:{R}\times {\mathcal {C}}\rightarrow {R}^n \) takes \({R}\times (\hbox {bounded sets of } {\mathcal {C}})\) into bounded sets of \({R}^n\), \(f(t, 0)=0\). Assume further that \(u,\upsilon ,\omega : {R}^+\rightarrow {R}^+\) are continuous nondecreasing functions, u(s) and \(\upsilon (s)\) are positive for \(s>0\), and \(u(0)=\upsilon (0)=0\). If there exists a continuous differential functional \(V:{R}\times {\mathcal {C}}\rightarrow {R}\) such that
then the equilibrium point \(x=0\) of the system (4) is stable. If \(\omega (s)>0\) for \(s>0\), then the equilibrium point \(x=0\) of the system (4) is asymptotically stable.
Proof
For any \(\varepsilon >0\), there exists a \(\delta =\delta (\varepsilon )\), \(0<\delta <\varepsilon \), such that \(\upsilon (\delta )< u(\varepsilon )\). Let \(x(t;t_0,\phi )\) be the state solution of state equation (4) satisfying the initial condition \(x_{t_0}=\phi \). The inequality (15) and Property 1 imply
That is
By the property of the function \(\omega \), it follows
If \(\Vert \phi \Vert < \delta \), then the inequalities (14), (15), (18) and the nondecreasing property of the function \(\upsilon \) imply
The nondecreasing property of the function u(s) and the inequality (19) imply \(|x(t; t_0, \phi )|<\varepsilon \), \(t\ge t_0\). The continuity of the solution \(x(t;t_0,\phi )\) on \([t_0-r, \infty )\) implies the stability of the equilibrium point \(x=0\) of the system (4).
To prove that the equilibrium point \(x=0\) of the system (4) is asymptotically stable, it is sufficient to prove the state solution \(x(t; t_0, \phi )\) of the system (4)–(5) satisfies: \(\lim \limits _{t\rightarrow +\infty }x(t; t_0, \phi )=0 \). Otherwise, if the equilibrium point \(x=0\) of the system (4) is not asymptotically stable, then there exist constants \(\eta >0\), \(t_i\in (t_0,+\infty )\), \(i=1,2,\ldots \), such that
where \(t_i\rightarrow \infty \) as \(i\rightarrow +\infty \). It follows from the inequality (15) that
That is
Since the function \(\omega \) is nondecreasing, we get by the inequality (20) that
Combining (22) with (23) yields
When \(t_i\) is large enough, it follows
This is a contradiction. This proves the asymptotical stability. The proof is completed.
Remark 1
Theorem 1 is an extension of Theorem 2.1 of the Ref. [29, Page 132].
Suppose that f: \(R\times {\mathcal {C}}\rightarrow R^n \) is continuous and \(f(t,0)=0\). Consider the RFDE(f)
subject to the initial condition
for any \(\sigma \in R\).
Theorem 2.1 of the Ref. [29], which is a stability criterion for the integer-order retarded functional differential equation (RFDE) (26), is as follows:
Suppose that f: \(R\times {\mathcal {C}}\rightarrow R^n \) takes \(R\times \) (bounded set of \({\mathcal {C}})\) into bounded sets of \(R^n\), and u, v, w: \(R^+\rightarrow R^+\) are continuous nondecreasing functions, u(s) and v(s) are positive for \(s>0\), and \(u(0)=v(0)=0\). If there is a continuous function V: \(R\times {\mathcal {C}}\rightarrow R\) such that
then the solution \(x=0\) of Eq. (26) is uniformly stable. If \(u(s)\rightarrow \infty \) as \(s\rightarrow \infty \), the solutions of Eq. (26) are uniformly bounded. If \(\omega (s)>0\) for \(s>0\), then the solution \(x=0\) of Eq. (26) is uniformly asymptotically stable.
It is worth to mention that the change of \(t_0\) yields the change of the system (4)–(5). Therefore, the uniform stability of the equilibrium point \(x=0\) of any fractional differential system with Caputo’s derivative including the system (4) does not hold.
By similar arguments used in the proof of Theorem 1, one can get the following theorem.
Theorem 2
Suppose that \(f:{R}\times {\mathcal {C}}\rightarrow {R}^n \) takes \({R}\times (\hbox {bounded sets of } {\mathcal {C}})\) into bounded sets of \({R}^n\) and \(f(t,0)=0\). Assume further that \(u,\upsilon ,\omega : {R}^+\rightarrow {R}^+\) are continuous nondecreasing functions, u(s) and \(\upsilon (s)\) are positive for \(s>0\), and \(u(0)=\upsilon (0)=0\). If there exists a continuous differential function \(V:{R}\times {R}^n\rightarrow {R}\) such that for \(t\in {R}\), \(x\in {R}^n\) and \(\varphi \in {\mathcal {C}}\),
then the equilibrium point \(x=0\) of the system (4) is stable. If \(\omega (s)>0\) for \(s>0\), then the equilibrium point \(x=0\) of the system (4) is asymptotically stable.
Remark 2
Theorem 2 also holds for \(\theta =0\). That is, Theorem 2 is valid for nonlinear fractional differential system
for \(t_0\in R^+\). If \(t_0=0\) and \(\theta =0\), then Theorem 2 is similar to theorems of the Ref. [26].
By Theorem 2 and Lemma 1, we can get the following corollary.
Corollary 1
Suppose that \(f:{R}\times {\mathcal {C}}\rightarrow {R}^n \) takes \({R}\times (\hbox {bounded sets of} \ {\mathcal {C}})\) into bounded sets of \({R}^n\) and \(f(t, 0)=0\). If for any \(x\in {R}^n\), \(x^\mathrm{T}(t)f(t, x_t)\le 0\) for \(t\ge t_0\), then the system (4) is stable. Furthermore, if for any \(x\ne 0\), it holds \(x^\mathrm{T}(t)f(t,x_t)<0\), then the system (4) is asymptotically stable.
Proof
Take a Lyapunov function \( V({t,x(t)}) = \frac{1}{2}x^\mathrm{T} (t)x(t)\). Lemma 1 implies
If \(x^\mathrm{T} (t)f({t,x_t }) \le 0\), then \({}^C D_{t_0,t}^\alpha V(t, x(t)) \le 0\). Theorem 2 implies the system (4) is stable.
If \(x^\mathrm{T} (t)f(t,x_t)< 0, \quad \forall x\ne 0\), then \({\hbox {}^C} D_{t_0,t}^\alpha V(t, x(t)) < 0\). Theorem 2 implies that the system (4) is asymptotically stable.
Remark 3
For the case \(\theta =0\), Corollary 1 also holds. Therefore, Corollary 1 here is an extension of Corollary 1 of the Ref. [27] and Theorem 3.1 of the Ref. [28]. In order to compare the Corollary 1 of this article with Corollary 1 of the Ref. [27] and Theorem 3.1 of the Ref. [28], we list them in the following.
Corollary 1 of the Ref. [27] is as follows:
For the fractional order system
where \(\alpha \in (0,1)\), \(x=0\) is the equilibrium point and \(x(t)\in R\), if the following condition is satisfied
then the origin of the system (33) is stable. And if
then the origin of the system (33) is asymptotically stable.
Theorem 3.1 of the Ref. [28] is as follows.
Consider the system
where \(0<\alpha <1\). \( \varOmega \subset R\) is a domain that contains the origin \(x=0\). Suppose further that \(f(x)\in C^1(\varOmega )\) with \(f(0)=0\). If \(x\cdot f(x)\le 0\), then the equilibrium point \(x=0\) is stable. Further, if \(x\ne 0\) implies \(x\cdot f(x)< 0\), then the equilibrium point \(x=0\) is asymptotically stable.
Corollary 2
Assume \(\alpha \in (0,1)\). Consider the fractional delayed differential system
where f, g is continuous and satisfy Lipschitz conditions in the second variable \(\varphi \in C([-r,0], {R}^n)\), \(f(t,0)=0\), \(\lim \limits _{x\rightarrow 0} \frac{|g(t,x_t)|}{|f(t,x_t)|}=0\). If \(x^\mathrm{T} f<0\) for \(x\ne 0\), then the system (37) is locally asymptotically stable.
Proof
Obviously, there exists a positive real number \(\delta >0\) (\(\delta \) may be small enough) such that if \(|x(t)|<\delta \), it holds \(\left| \frac{f(t,x_t)}{g(t,x_t)}\right| <\frac{1}{2}\). In consequence, \(|x(t)|<\delta \) and \(x^\mathrm{T} f<0\) imply
By Corollary 1, the system (37) is locally asymptotically stable.
The following theorems are Razumikhin-type stability criteria for the fractional delayed differential system (4).
Theorem 3
Suppose that \(f:{R}\times {\mathcal {C}}\rightarrow {R}^n \) takes \({R}\times (\hbox {bounded sets of}\ {\mathcal {C}})\) into bounded sets of \({R}^n\), \(f(t, 0)=0\). Assume further that \(u,\upsilon ,\omega : {R}^+\rightarrow {R}^+\) are continuous nondecreasing functions, u(s) and \(\upsilon (s)\) are positive for \(s>0\), and \(u(0)=\upsilon (0)=0\). If there exists a continuous function \(V:{R} \times {R}^n \rightarrow {R}\) such that for \(t\in {R}\), \(x\in {R}^n\) and \(\varphi \in {\mathcal {C}}\),
and \(V(t+\theta , \varphi (\theta ))\le V(t, \varphi (0))\) implies
for \(\theta \in [-r,0]\), then the equilibrium point \(x=0\) of the system (4) is stable.
Proof
For \( t\in [t_0, \infty ) \), \( \varphi \in {\mathcal {C}} \), define
Then there is a \(\theta _0(t)\in [-r, 0]\) such that \({\overline{V}}(t, \varphi ) = V(t+\theta _0, \varphi (\theta _0))\). It follows \(V(t+\theta , \varphi (\theta ))\le V(t+\theta _0, \varphi (\theta _0))\) for \(\theta \in [-r, 0]\). If \(\theta _0<0\), then for \(h>0\) sufficiently small
and therefore \(\dot{{\overline{V}}}(t,\varphi )=0\). Thus it follows that
If \(\theta _0=0\), then \({\overline{V}}(t, \varphi )=V(t, \varphi (0))\). We have from the condition (40) that
By similar arguments used in the proof of Theorem 1, it follows that the equilibrium point \(x=0\) of the system (4) is stable. This completes the proof.
Remark 4
Theorem 3 is parallel to Theorem 4.1 of the Ref. [29, P152], which is as follows:
Suppose f:\(R\times {\mathcal {C}}\rightarrow R^n \) takes \(R\times \) (bounded sets of \({\mathcal {C}}\)) into bounded sets of \(R^n\) and suppose further that \(u,v, w:R^+\rightarrow R^+ \) are continuous, nondecreasing functions, u(s), v(s) positive for \(s>0\), \(u(0)=v(0)=0\), v strictly increasing. If there is a continuous function \(V:R\times R^n \rightarrow R \) such that
and
for \(\theta \in [-r,0]\), then the solution \(x=0\) of Eq. (26) is uniformly stable.
Theorem 4
Suppose that all the conditions of Theorem 3 are satisfied. Assume further that the function \(\upsilon \) is strictly increasing. Namely, \(\upsilon \in {\mathcal {K}}\). If there is a continuous nondecreasing function \(p(s)>s\) for \(s>0\) such that the condition (40) is strengthened to
for \(t\ge t_0\) and \(\theta \in [-r,0]\), then the equilibrium point \(x=0\) of the system (4) is asymptotically stable.
Proof
By Theorem 3, the equilibrium point \(x=0\) of the system (4) is stable.
The fact that the solution \(x=0\) of the system (4) is stable shows that for the stability constant \(\varepsilon _0=1\), we can choose a constant \(\delta _0>0\) such that for \(\Vert \phi \Vert \le \delta _0\), the state solution \(|x(t; t_0,\phi )|\) of the system (4)–(5) satisfying \(|x(t;t_0,\phi )|<\varepsilon _0=1\). In the remainder of the proof, we denote the state solution \(x(t;t_0,\phi )\) of the system (4)–(5) as x(t) for convenience.
In order to prove that the equilibrium point \(x=0\) of the system (4) is asymptotical stability, it is sufficient to prove that, for any \(\eta >0\) and \(u(\eta )<v(\varepsilon _0)=v(1)\), there exists a \(T_0\ge t_0\) such that \(V(t,x(t))<u(\eta )\) on \([T_0, \infty )\).
From the properties of the continuous function p(s), there exists a number \(a>0\) such that \(p(s)-s>a\) on the interval \([u(\eta ), v(1)]\). Let \(n_0\) be the first positive integer such that \(u(\eta )+n_0 a \ge v(1)\). We first prove that when t is large enough, \(V(t,x(t))\le u(\eta )+(n_0-1)a.\) Otherwise, there exist constants \(t_i>t_0\), \(i=1,2, \ldots \), satisfying \(t_i\rightarrow \infty \) as \(i\rightarrow \infty \) such that
It follows that
By the condition (39) and (49), we have
From (51), we have
Since v is class-K function, we have
By the fact that the function \(\omega \) is continuous and nondecreasing, it follows
\(t_i\rightarrow \infty \) will yield \(V(t,x(t))<0\). This is a contraction. In consequence, \(V(t,x(t))\le u(\eta )+(n_0-1)a\) when t is large enough. Without loss of generality, we assume that there exists a constant \(T_1>t_0\) such that
We now prove that there exists a constant \(T_2>T_1+r\) such that
Otherwise, there exist constants \(t_{m}>T_1+r\), \(m=1,2,\ldots \), \(t_{m}\rightarrow \infty \) as \(m\rightarrow \infty \) such that
It then follows from (59) that
Combining (57) with (60) yields
By (63), if \(t_{m}\rightarrow \infty \), then \(V(t,x(t))<0\). This is a contraction. In consequence, we can choose a constant \(T_2>T_1+r\) such that
Repeat above process \(n_0\) times, we can choose a constant \(T_{n_0}>t_0\) such that
From (65) and the condition (39), it follows
It then follows \(|x(t)|<\eta \) on \([T_{n_0}, +\infty )\). This shows that the equilibrium point \(x=0\) of the system (4) is asymptotically stable. The proof is completed.
Remark 5
Theorem 4 is parallel to Theorem 4.2 of the Ref. [29, P152] which is as follows:
Suppose that all of the conditions of Theorem 4.1 are satisfied and in addition \(\omega (s)>0\) for \(s>0\). If there is a continuous nondecreasing function \(p(s)>s\) for \(s>0\) such that the condition (46) is strengthened to
for \(\theta \in [-r,0]\), then the solution \(x=0\) of the RFDE(f) (26) is uniformly asymptotically stable. If \(u(s)\rightarrow \infty \) as \(s\rightarrow \infty \), then the solution \(x=0\) is also a globally attractor for the RFDE(f).
4 Illustrative examples and numerical simulations
In this section, we present four examples to illustrate the applications of our results. All numerical simulations are based on the predictor–corrector algorithm [33].
Example 1
Consider the system
We first check the system (68) with a initial state \(x_{0}(\theta )=\phi (\theta )\) has only unique solution on \([0, \infty )\). Here the initial state \(\phi (\theta )\) is continuous for \(\theta \in [-1,0]\). Denote \(f(t,x_t)=-x(t)x^2(t-1)\). For \(t\in [0,1]\), the system (68) is equivalent to
For any \(x_t, y_t\in C([-1,0], R)\), \(t\in [0, 1]\), we have
where the Lipschitz constant \(L_1=\max \limits _{t\in [0,1]}|\phi ^2(t-1)|\). In consequence, the functional \(f(t,x_t)=-x(t)\cdot x^2(t-1)\) satisfies Lipschitz condition with a Lipschitz constant \(L_1\). By Lemma 2, the system (68) with the initial state \(x_0(\theta )=\phi (\theta )\) has a unique solution on [0, 1]. Without loss of generality, we denote \(x(t;0,\phi (\theta ))=\phi _{1}(t)\) defined on [0, 1] the state solution of the system (68) with the initial state \(\phi (\theta )\).
For \(t\in [1,2]\), the system (68) is equivalent to
For any \(x_t,y_t\in C([-1,0], R)\), \(t\in [1,2]\), we have
where \(L_2=\max \limits _{t\in [1,2]}\phi _1^2(t-1)\). Take \(L_{1,2}=\max \{L_1,L_2\}\). For any \(x_t, y_t\in {\mathcal {C}}\), \(t\in [0, 2]\), combining (70) with (72) yields
Therefore, the functional \(f(t,x_t)=-x(t)\cdot x^2(t-1)\) in the system (68) satisfied Lipschitz condition on [0, 2] with a Lipschitz constant \(L_{1,2}\). By Lemma 2, the system (68) with the initial state \(\phi (\theta )\) has a unique solution on [0, 2].
Similarly, we can show that the system (68) with the initial state \(\phi (\theta )\) has a unique state solution on [0, 3], \([0,4], \ldots \). By mathematical induction, it follows that the system (68) with the initial state \(\phi (\theta )\) has a unique solution \(x(t;0,\phi (\theta ))\) on \([0, \infty )\).
Obviously, \(x(t)f(t,x_t)=-x^2(t)x^2(t-1)\le 0\). By Corollary 1, the system (68) is stable. The numerical simulation of the state solution \(x(t;0,\phi (\theta ))\) of the system (68) with the initial state \(\phi (\theta )=1+\theta \), \(\phi (\theta )=0.5(1+\theta )\), respectively, for \(\theta \in [-1, 0]\) is given in Fig. 1.
On the one hand, it can be seen from Fig. 1 that the initial states satisfy \(\Vert 0.5(1+\theta )\Vert < \Vert 1+\theta \Vert \) for \(\theta \in [-1, 0]\) and the state solutions satisfy \(\Vert x(t;0, 0.5(1+\theta ))\Vert < \Vert x(t; 0, 1+\theta )\Vert \) for \(t\in [0, \infty )\). This fact coincides with the stability of the system (68).
On the other hand, the state solutions \(x(t;0, 1+\theta )\) and \(x(t; 0, 0.5(1+\theta ))\) in Fig. 1 do not tend to 0 as \(t \rightarrow \infty \). This does not show that the system (68) is asymptotically stable.
Example 2
Consider the system
Just as the existence and uniqueness of the global state solution of the system (68) with a initial state is checked, it is easy to check that the system (74) with the initial state \(\phi (\theta )\) for \(\theta \in [-1, 0]\) has a unique state solution \(x(t; 0, \phi (\theta ))\) on \([0,\infty )\).
Denote \(f(t,x_t)=-x(t)\left[ 1+x^2(t-1)\right] \). Obviously,
for \(x\ne 0\). Corollary 1 implies that the system (74) is asymptotically stable. The numerical simulation of the state solution of the system (74) with the initial state \(\phi (\theta )=1+\theta \) for \(\theta \in [-1, 0]\) is given in Fig. 2.
From Fig. 2, the state solution \(x(t;0, 1+\theta )\rightarrow 0\) as \(t\rightarrow \infty \). This shows that the state solution \(x(t;0, 1+\theta )\) is attractive, which coincides with the asymptotical stability of the system (74).
Example 3
Consider the system
By Lemma 2, it is easy to check that the system (76) with a initial state \(x_0(\theta )=\phi (\theta )\) has a unique solution on \([0, \infty )\).
Denote \(f(t,x_t)=-x(t) \left[ 1+x^2(t-1)\right] \), \(g(t,x_t)=\pm 2x(t)x^3(t-1)\). Then we can check that \(\lim \limits _{x\rightarrow 0}\frac{g(t,x_t)}{f(t,x_t)}=0\). Obviously, \(x(t)f(t,x_t)<0\) for \(x\ne 0\). Corollary 2 implies that the system (76) is locally asymptotically stable. The numerical simulation of the system (76) with the initial state \(x_0(\theta )=0.1(1+\theta )\) is given in Fig. 3.
From Fig. 3, it can be seen that the state solution \(x(t; 0, 0.1(1+\theta ))\) of the system (76) tends to be 0. This coincides with the local stability of equilibrium point \(x=0\) of the system (76).
Remark 6
Denote \(F(t,x_t)=-x(t)\left[ 1+x^2(t-1)\right] \pm 2x(t) x^3(t-1)\). If follows
Corollary 1 is invalid to the system (76). However, Corollary 2 implies that the system (76) is locally asymptotically stable. This shows that Corollary 2 is sometimes suitable for application.
Example 4
Consider the system
where \(\alpha \in (0, 1)\). We first check that the functional \(f(t,x_t)=-2x(t)+x(t-1)\) satisfies Lipschitz condition. In fact, for any \(x_t, y_t\in C([-1,0], R)\) and \(t\in R^+\), we have
which implies that the functional \(f(t,x_t)=-2x(t)+x(t-1)\) satisfies Lipschitz condition with a Lipschitz constant \(L=3\). By Lemma 3, the system (78) with the initial state \(\phi (\theta )\in C([-1,0], R)\) has a unique state solution on \([0, \infty )\).
Choose a constant \(q=\frac{3}{2}\). When \(q|x(t)|>|x(t-1)|\), we get
for \(x\ne 0\). By Theorem 4, the equilibrium point \(x=0\) of the system (78) is asymptotically stable.
The numerical simulations of the state solution of the system (78) with \(\alpha =1, 0.9, 0.5\) and \(\phi (\theta )=\theta ^2\) for \(\theta \in [-1, 0]\) is given in Fig. 4.
From Fig. 4, it can be seen that no matter \(\alpha =1, 0.9\) or \(\alpha =0.5\), the state solution \(x(t; 0, \theta ^2 )\rightarrow 0\) as \(t\rightarrow \infty \). This coincides with the fact that the equilibrium point \(x=0\) of the system (78) is asymptotically stable.
5 Conclusions
In this paper, Lyapunov functional and Lyapunov function are used, respectively, to investigate the stability of the fractional delayed differential systems. Some stability criteria including the two Razumikhin-type stability theorems are derived. Four examples are given for illustrations. Moreover, numerical simulations are provided to show the validity of the results.
References
Podlubny, I.: Fractional Differential Equations. Academic Press, New York (1993)
Miller, K.S., Ross, B.: An Introduction to the Fractional Calculus and Differential Equations. Wiley, New York (1993)
Kilbas, A.A., Srivastava, H.M., Trujillo, J.J.: Theory and Applications of Fractional Differential Equations. Elsevier, Amsterdan (2006)
Lakshmikantham, V., Leela, S., Vasundhara Devi, J.: Theory of Fractional Dynamic Systems. Cambridge Scientific Publishers, Cambridge (2009)
Diethelm, K.: The Analysis of Fractional Differential Equations. Springer, Heidelberg (2010)
Zhou, Y.: Basic Theorem of Fractional Differential Equations. World Scientific, Singapore (2014)
Chauhan, A., Dabas, J.: Local and global existence of mild solution to an impulsive fractional functional integro-differential equation with nonlocal condition. Commun. Nonlinear Sci. Numer. Simul. 19, 821–829 (2014)
Benchohra, M., Henderson, J., Ntouyas, S.K., Ouahab, A.: Existence results for fractional order functional differential equations with infinite delay. J. Math. Anal. Appl. 338, 1340–1350 (2008)
Zhou, Y., Jiao, F., Li, J.: Existence and uniqueness for fractional neutral differential equations with infinite delay. Nonlinear Anal. 71, 3249–3265 (2009)
Agarwal, R.P., Zhou, Y., He, Y.: Existence of fractional neutral functional differential equations. Comput. Math. Appl. 59, 1095–1100 (2010)
Wang, J.R., Zhou, Y.: Existence of mild solutions for fractional delay evolution systems. Appl. Math. Comput. 218, 357–367 (2011)
Deng, J., Deng, Z., Yang, L.: An existence and uniqueness result for fractional order semilinear functional differential equations with nondense domain. Appl. Math. Lett. 24, 924–928 (2011)
Yang, Z., Cao, J.: Initial value problems for arbitrary order fractional differential equations with delay. Commun. Nonlinear Sci. Numer. Simul. 18, 2993–3005 (2013)
Jankowski, T.: Existence results to delay fractional differential equations with nonlinear boundary conditions. Appl. Math. Comput. 219, 9155–9164 (2013)
Chen, Y., Moore, K.L.: Analytical stability bounded for a class of delayed fractional-order dynamic systems. Nonlinear Dyn. 29, 191–200 (2002)
Lazarević, M.P.: Finite time stability analysis of \(PD^{\alpha }\) fractional control of robotic time-delay systems. Mech. Res. Commun. 33, 269–279 (2006)
Kaslik, E., Sivasundaram, S.: Analytical and numerical methods for the stability analysis of linear fractional delay differential equation. J. Comput. Appl. Math. 236, 4027–4041 (2012)
Mesbahi, A., Haeri, M.: Stability of linear ime invariant fractional delay systems of retardee type in the space of delay parameters. Automatica 49, 1287–1294 (2013)
Lakshmikantham, V.: Theorem of fractional functional differential equations. Nonlinear Anal. 69, 3337–3343 (2008)
Liu, S., Li, X., Jiang, W., et al.: Mittag–Leffler stability of nonlinear fractional neutral singular systems. Commun. Nonlinear Sci. Numer. Simul. 17, 3961–3966 (2012)
Zhou, Y., Jiao, F., Pecaric, J.: On the Cauchy problem for fractional functional differential equations in Banach spaces. Topol. Methods Nonlinear Anal. 42, 119–136 (2013)
Bhallekar, S.B.: Stability analysis of a class of fractional delay differential equations. J. Phys. 81(2), 215–224 (2013)
Gao, Z.: A graphic stability criterion for non-commensurate fractional-order time-delay systems. Nonlinear Dyn. 78, 2101–2111 (2014)
Stamova, I.: Global Mittag–Leffler stability and synchronization of impulsive fractional-order neural networks with time-varying delays. Nonlinear Dyn. 77, 1251–1260 (2014)
Hu, J.-B., Lu, G.-P., Zhang, S.-B., Zhao, L.-D.: Lyapunov stability theorem about fractional system without and with delay. Commun. Nonlinear Sci. Numer. Simul. 20, 905–913 (2015)
Li, Y., Chen, Y., Pudlubny, I.: Stability of fractional-order nonlinear dynamic systems: Lyapunov direct method and generalized Mittag-Leffler stability. Comput. Math. Appl. 59, 1810–1821 (2010)
Aguila-Camacho, N., Duarte-Mermoud, M.A., Gallegos, J.A.: Lyapunov functions for fractional order systems. Commun. Nonlinear Sci. Numer. Simul. 19, 2951–2957 (2014)
Zhou, X.-F., Hu, L.-G., Liu, S., et al.: Stability criterion for a class of nonlinear fractional differential systems. Appl. Math. Lett. 28, 25–29 (2014)
Hale, J.K., Verduyn Lunel, S.M.: Introduction to Functional Differential Equations. Springer, Berlin (1993)
Yu, L., Fei, S.: Robustly stable switching neural control of robotic manipulators using average dwell-time approach. Trans. Inst. Meas. Control 36, 789–796 (2014)
Yu, L., Fei, S., Sun, L., et al.: Design of robust adaptive neural switching controller for robotic manipulators with uncertainty and disturbances. J. Intell. Robot. Syst. 77, 571–581 (2015)
Yu, L., Fei, S., Sun, L., et al.: Fuzzy approximation of a novel nonlinear adaptive switching controller design. Circuits Syst. Signal Process. 34, 377–391 (2015)
Diethelm, K., Ford, N.J., Freed, A.: A predictor–corrector approach for the numerical solution of fractional differential equations. Nonlinear Dyn. 29, 3–22 (2002)
Acknowledgments
The authors would like to thank the reviewers for their valuable comments and suggestions, which improve the quality of this paper. This paper is supported by National Natural Science Foundation of China (11371027, 11471015), Natural Science Foundation of Anhui Province (1508085MA01), Research Fund for Doctoral Program of Higher Education of China (20123401120001, 20133401120013), Natural Science Foundation for Colleges and Universities of Anhui Province (KJ2012A032) and The Program of Academic Innovation Research for Postgraduate of Anhui University (yfc100013).
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Wen, Y., Zhou, XF., Zhang, Z. et al. Lyapunov method for nonlinear fractional differential systems with delay. Nonlinear Dyn 82, 1015–1025 (2015). https://doi.org/10.1007/s11071-015-2214-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11071-015-2214-y