This paper investigates impulsive synchronization of multilinks delayed coupled complex networks with perturb effects. Based on the comparison theory of impulsive differential system, a novel synchronization criterion is derived and an impulsive controller is designed simultaneously. Finally, numerical simulations demonstrate the effectiveness of the proposed synchronization criteria.
1. Introduction
In the past few decades, the problem of control and synchronization of complex dynamical networks has been extensively investigated in various fields of science and engineering due to its many potential practical applications [1–8]. One important consideration in practical networks is the existence of time delays because obstructions to the transmission of signals are inevitable in a biological neural network, in an epidemiological model, in a communications network, or in an electrical power grid. Since recently, there are many studies on dynamical networks with time delays [9–16]. Moreover, the multidelayed coupling consists of providing more information about the dynamics in nodes to the other nodes in the network, such as the transportation network; we all know that the transmission speed is different among highway network railway network, and airline network. So we can use timedelay to describe these networks [17]. In [17], the authors studied synchronization of a class of timedelayed complex dynamical networks with multilinks, and this model is suitable to investigate and simulate more realistic complex networks, so we should pay attention to this network with multilinks. Moreover, in [9–11, 17–20], the authors studied synchronization of complex network with time-varying coupling delay by designing a controller and adaptive updated laws. However, the controller designed for such an adaptive synchronization is usually quite complex, it will be useful to find a simple structure to solve this problem, the impulsive controller seems to have a simple structure, and impulsive control is an artificial control strategy which is cheaper to operate compared with other control strategy. Motivated by the above discussions, we investigate impulsive synchronization for such a complex networks model in this paper, and the novel synchronization criterion is derived.
The rest of this work is organized as follows. Section 2 gives the problem formulation. Section 3 gives synchronization scheme. Section 4 gives illustrative example. Section 5 gives the conclusion of the paper.
2. Problem Formulation
In [17], the authors achieve synchronization between two complex networks with multilinks by designing effective controller. For simplicity, the complex network model is written in the following form:
(2.1)x˙i(t)=f(xi(t))+∑l=0m-1∑j=1Naijlxj(t-τl)=f(xi(t))+∑j=1Naij0xj(t)+∑j=1Naij1xj(t-τ1)+⋯+∑j=1Naijm-1xj(t-τm-1),
where xi=(x1,x2,…,xn)T∈Rn, f:Rn→Rn standing for the activity of an individual subsystem is a vector value function. Al=(aijl)N×N∈RN×N(l=0,1,…,m-1) is the lth subnetwork’s topological structure. The definition of aijl is that in the lth sub-network, if there exists a link from node i to j(i≠j), then aijl≠0. Otherwise, aijl=0·τl(l=0,1,…,m-1) is time-delay of the lth subnetwork compared to the zero subnetwork (τ0=0) which is without time delayed.
Remark 2.1.
In [17], aiil=-∑j=1,i≠jNaijl is defined, we are not concerned whether the coupling matrix Al satisfies aiil=-∑j=1,i≠jNaijl in this paper.
In the paper, we have the following mathematical preliminaries.
Assumption 2.2.
We assume that f(xi(t)) is Lipschitz continuous on xi(t), that is, there exists a positive constant η>0 such that
(2.2)|f(yi(t))-f(xi(t))|≤ηei(t),∀xi(t),yi(t)∈Rn.
Assumption 2.3.
We also assume that σ(xi(t)) is Lipschitz continuous on xi(t), and one can consider
(2.3)σ(yi(t),t)-σ(xi(t),t)=M(xi(t),yi(t))ei(t),
where ∥M(xi(t),yi(t))∥≤H,H>0.
3. Synchronization Scheme
In this section, we will investigate impulsive synchronization of the complex networks with perturb functions. The multidelayed coupled complex network with perturb functions can be described by(3.1)x˙i=f(xi(t))+∑j=1Naij0xj(t)+∑j=1Naij1xj(t-τ1)+⋯+∑j=1Naijm-1xj(t-τm-1)+σ(xi(t),t).
We take the network given by (3.1) as the driving network and a response network with impulsive control scheme which is given by
(3.2)y˙i=f(yi(t))+∑j=1Naij0yj(t)+∑j=1Naij1yj(t-τ1)+⋯+∑j=1Naijm-1yj(t-τm-1)+σ(yi(t),t),t≠tk,(3.3)Δyi=yi(tk+)-yi(tk-)=Bik(yi-xi),t=tk,
where y=(y1,y2,…,yn)T∈Rn.
Let ei(t)=yi(t)-xi(t), then we have the following error system:
(3.4)e˙i=f(yi(t))-f(xi(t))+∑j=1Naij0(yj(t)-xj(t))+∑j=1Naij1(yj(t-τ1)-xj(t-τ1))+⋯+∑j=1Naijm-1(yj(t-τm-1)-xj(t-τm-1))+M(xi(t),yi(t))ei,t≠tk,(3.5)Δei=ei(tk+)-ei(tk-)=Bikei,t=tk,
where e(tk+)=limt→tk+e(t), e(tk)=limt→tk-e(t)=e(tk-).
Theorem 3.1.
Let Assumptions 2.2–2.3 hold, αr-1=max(aijr-1)2, r=1,2,…,m. If there exists a constant θ≥1 such that
(3.6)lnθρk+2((η+∑r=1mαr-1+m+H)(tk+1-tk))≤0,
then the driving network (3.1) and the response network (3.2) can realize impulsive synchronization.
Proof.
We choose a nonnegative function as
(3.7)V(t)=12∑i=1N(ei(t))Tei(t)+∑i=1N∫t-τ1t(ei(s))Tei(s)ds+⋯+∑i=1N∫t-τm-1t(ei(s))Tei(s)ds.
Then the differentiation of V along the trajectories of (3.4) is
(3.8)V˙(t)=∑i=1N(ei(t))T[(f(yi(t))-f(xi(t)))+∑j=1Naij0(yj(t)-xj(t))+∑j=1Naij1(yj(t-τ1)-xj(t-τ1))+⋯+∑j=1Naijm-1(yj(t-τm-1)-xj(t-τm-1))]+∑i=1N[(ei(t))Tei(t)-(ei(t-τ1))Tei(t-τ1)]+⋯+∑i=1N[(ei(t))Tei(t)-(ei(t-τm-1))Tei(t-τm-1)]+∑i=1N(ei(t))TM(xi(t),yi(t))ei(t)≤∑i=1Nη(ei(t))Tei(t)+∑i=1N∑j=1Naij0(ei(t))Tej(t)+∑i=1N∑j=1Naij1(ei(t))Tej(t-τ1)+⋯+∑i=1N∑j=1Naijm-1(ei(t))Tej(t-τm-1)+∑i=1N[(ei(t))Tei(t)-(ei(t-τ1))Tei(t-τ1)]+⋯+∑i=1N[(ei(t))Tei(t)-(ei(t-τm-1))Tei(t-τm-1)]+∑i=1NH(ei(t))Tei(t)≤∑i=1Nη(ei(t))Tei(t)+∑i=1N∑j=1N(aij0)2(ei(t))Tei(t)+∑j=1N(ej(t))Tej(t)+∑i=1N∑j=1N(aij1)2(ei(t))Tei(t)+∑j=1N(ej(t-τ1))Tej(t-τ1)+⋯+∑i=1N∑j=1N(aijm-1)2(ei(t))Tei(t)+∑j=1N(ej(t-τm-1))Tej(t-τm-1)+∑i=1N[(ei(t))Tei(t)-(ei(t-τ1))Tei(t-τ1)]+⋯+∑i=1N[(ei(t))Tei(t)-(ei(t-τm-1))Tei(t-τm-1)]+∑i=1NH(ei(t))Tei(t)=∑i=1Nη(ei(t))Tei(t)+∑i=1N∑j=1N(aij0)(ei(t))Tei(t)+∑j=1N(ej(t))Tej(t)+∑i=1N∑j=1N(aij1)2(ei(t))Tei(t)+⋯+∑i=1N∑j=1N(aijm-1)2(ei(t))Tei(t)+∑i=1N(ei(t))Tei(t)+⋯+∑i=1N(ei(t))Tei(t)+∑i=1NH(ei(t))Tei(t)≤∑i=1Nη(ei(t))Tei(t)+∑i=1N∑j=1N(aij0)2(ei(t))Tei(t)+∑j=1N(ej(t))Tej(t)+∑i=1N∑j=1N(aij1)2(ei(t))Tei(t)+⋯+∑i=1N∑j=1N(aijm-1)2(ei(t))Tei(t)+∑i=1N(ei(t))Tei(t)+⋯+∑i=1N(ei(t))Tei(t)+∑i=1NH(ei(t))Tei(t)=(η+∑r=1mαr-1+m+H)∑i=1N(ei(t))Tei(t)≤2(η+∑r=1mαr-1+m+H){12∑i=1N(ei(t))Tei(t)+∑i=1N∫t-τ1t(ei(s))Tei(s)ds+⋯+∑i=1N∫t-τm-1t(ei(s))Tei(s)ds}=2(η+∑r=1mαr-1+m+H)V(t).
This implies that
(3.9)V(e(t))≤V(e(tk-1+))exp(2(η+∑r=1mαr-1+m+H)(t-tk-1)),t∈(tk-1,tk],k=1,2,…
On the other hand, when t=tk, we have
(3.10)V(e(t+))=12∑i=1N(ei(tk))T[I+Bik]T[I+Bik]ei(tk)+∑i=1N∫t-τjt(ei(s))T[I+Bik]T[I+Bik]ei(s)ds+∑i=1N∫t-djt(ei(s))T[I+Bik]T[I+Bik]ei(s)ds≤λmax[(I+Bik)T(I+Bik)]{12∑i=1N(ei(t))Tei(t)+∑i=1N∫t-τ1t(ei(s))Tei(s)ds≤λmax[(I+Bik)T(I+Bik)]+⋯+∑i=1N∫t-τm-1t(ei(s))Tei(s)ds}≤λmax[(I+Bik)T(I+B)ik]{12∑i=1N(ei(t))Te(t)+∑i=1N∫t-τ1t(ei(s))Tei(s)ds≤λmax[(I+Bik)T]+⋯+∑i=1N∫t-τm-1t(ei(s))Tei(s)ds}=ρkV(e(tk)),
where ρk=λmax[(I+Bik)T(I+Bik)].
When t∈(t0,t1], V(e(t))≤V(e(t0+))exp(2(η+∑r=1mαr-1+m+H)(t-t0)), then
(3.11)V(e(t1))≤V(e(t0+))exp(2(η+∑r=1mαr-1+m+H)(t1-t0)).
So,
(3.12)V(e(t1+))≤ρ1V(e(t1))≤ρ1V(e(t0+))exp(2(η+∑r=1mαr-1+m+H)(t1-t0)).
In the same way, for t∈(t1,t2], we have
(3.13)V(e(t))≤V(e(t1+))exp(2(η+∑r=1mαr-1+m+H)(t-t1))≤ρ1V(e(t0+))exp(2(η+∑r=1mαr-1+m+H)(t-t0)).
In general for any t∈(tk,tk+1], one finds that
(3.14)V(e(t))≤V(e(t0+))ρ1ρ2⋯ρkexp(2(η+∑r=1mαr-1+m+H)(t-t0))
Thus for all t∈(tk,tk+1], k=1,2,…, we have
(3.15)V(e(t))≤V(e(t0+))ρ1ρ2⋯ρkexp(2(η+∑r=1mαr-1+m+H)(t-t0))≤V(e(t0+))ρ1ρ2⋯ρkexp(2(η+∑r=1mαr-1+m+H)(tk+1-t0))=V(e(t0+))ρ1exp(2(η+∑r=1mαr-1+m+H)(t2-t1))ρ2×exp(2(η+∑r=1mαr-1+m+H)(t3-t2))⋯ρkexp(2(η+∑r=1mαr-1+m+H)(tk+1-tk))×exp(2(η+∑r=1mαr-1+m+H)(t-t0)).
From the assumptions given in the theorem
(3.16)ρkexp(2(η+∑r=1mαr-1+m+H)(tk+1-tk))≤1θ,k=1,2,…
we have V(e(t))≤V(e(t0+))(1/θk)exp(2(η+∑r=1mαr-1+m+H)(t-t0)). That is V(e(t))≤V(e(t0+))(1/θk)exp(2(η+∑r=1mαr-1+m+H)(t-t0)),t≥t0.
When θ≥1, from [21], this implies that the origin in system (3.4) is globally asymptotically stable or the driving network is synchronized with the response network asymptotically for any initial conditions. This completes the proof.
Remark 3.2.
Systems (3.1)-(3.2) are the time-invariant complex networks. As discussed in [22–24], systems (3.1)-(3.2) are the time-varying complex networks, which is a more complicated research issue.
Remark 3.3.
Normally, it is difficult to control a complex networks by adding the controllers to all nodes, so it would be much better to use the pinning control method since the most complex networks have large number of nodes [25]. Regarding for the pinning control of the network systems (3.1)-(3.2), are the next research topic for us.
Remark 3.4.
For the transportation network, we all know that the transmission speed is different among highway network, railway network and airline network. So we can use multilinks delayed to describe these networks [17]. Also impulsive control is an artificial control strategy which is cheaper to operate compared with other control strategy, so impulsive control method of the network systems (3.1)-(3.2) should have potential applications.
4. Illustrative Example
It is well known that the Lorenz system families are typical chaotic systems and the Lü chaotic system is a member of the families which is known as [26]
(4.1)s˙=(-aa00c000-b)(s1s2s3)+(0-s1s3s1s2)def=Ps+W(s),
when a=36,c=20,b=3.
It is well known that the Lü attractor is bounded. Here we suppose that all nodes are running in the given bounded region. Our numerical analyses show that there exist constants M1=25,M2=30,M3=45.
Satisfying ∥yp∥,∥zp∥≤Mp for 1≤p≤3. Therefore, one has
(4.2)∥W(y)-W(z)∥≤(-y3(y1-z1)-z1(y3-z3))2+(y2(y1-z1)+z1(y2-z2))2≤2M12+M22+M32∥y-z∥≈64.6142∥y-z∥.
Obviously, ∥P∥≈52.9843. Thus the Lü system satisfies Assumption 2.2,η=117.5985. In the same way, it can be seen that the Chen system, the Lorenz system, the unified chaotic system and the Lorenz system families also satisfy Assumption 2.2. So, in the simulations, we select the Lü chaotic system as an example to show the effectiveness of the proposed method [27, 28].
According to Section 3, we show that the network with 4 nodes described by
(4.3)x˙i(t)=f(xi(t))+∑j=14aij0xj(t)+∑j=14aij1xj(t-τ1)+∑j=14aij2xj(t-τ2)+0.1xi(t),y˙i(t)=f(yi(t))+∑j=14aij0yj(t)+∑j=14aij1yj(t-τ1)+∑j=14aij2yj(t-τ2)+0.1yi(t),t≠tk,Δyi=yi(tk+)-yi(tk-)=Bik(yi-xi),t=tk.
In numerical simulation, let
(4.4)A0=(5-4-204-43-123-400-13-1),A1=(-33-101-45-121-200-302),A2=(-13-102-45-111-200-302),
we choose τ1=0.1, τ2=0.2, and the gain matrixes Bik(k=1,2,…) as a constant matrix, Bik=B=diag(-0.7,-0.8,-0.9), then ρk=0.09. Let θ=1.1, from lnθρk+2(η+∑r=1mαr-1+m+H)(tk+1-tk)≤0, then tk+1-tk≤0.0085, we let tk+1-tk=0.008. All initial values are x1i=1+0.5i, x2i=2+0.7i, x3i=2+0.8i, y1i=1-0.6i, y2i=2-0.8i, y3i=3-0.8i. Figure 1 shows the variance of the synchronization errors. We introduce the quantity E(t)=∑i=1N∥yi(t)-xi(t)∥2/N [29] which is used to measure the quality of the control process. It is obvious that when E(t) no longer increases, two networks achieve synchronization.
Synchronization errors with time t.
5. Conclusion
This paper deals with the problem of impulsive synchronization of multilinks delayed coupled complex networks with perturb effects. On the basis of the comparison theory of impulsive differential system, the novel synchronization criteriion is derived and an impulsive controller is designed simultaneously. Finally, numerical simulations are presented to verify the effectiveness of the proposed synchronization criteria.
Acknowledgments
The authors would like to thank the referees and the editor for their valuable comments and suggestions. This research is supported by the National Natural Science Foundation of China (61075060), the Science and Technology Research Key Program for the Education Department of Hubei Province of China (D20105001 and D20126002), the Excellent Young Key Talents Cultivation Plan of Hubei Province, China Postdoctoral Science Foundation (2012M511663), the Research Fund for the Technology Development Program of Higher School of Tianjin (20111004), the Social Science Planning Project of Tianjin (TJLJ11-009), and the young talents Fund of Anhui Provinc of China (2011SQRL164).
LüJ.YuX.ChenG.ChengD.Characterizing the synchronizability of small-world dynamical networks200451478779610.1109/TCSI.2004.8236722066214LiX.ChenG.Synchronization and desynchronization of complex dynamical networks: an engineering viewpoint200350111381139010.1109/TCSI.2003.8186112024565XuY.ZhouW.FangJ.SunW.Adaptive lag synchronization and parameters adaptive lag identification of chaotic systems201037434344134462-s2.0-7795556894110.1016/j.physleta.2010.06.064ZhouJ.XiangL.LiuZ.Global synchronization in general complex delayed dynamical networks and its applications2007385272974210.1016/j.physa.2007.07.0062584888PengH.LiL.YangY.SunF.Conditions of parameter identification from time series2011833803620210.1103/PhysRevE.83.036202ShenB.WangZ.ShuH.WeiG.On nonlinear H∞ filtering for discrete-time stochastic systems with missing measurements20085392170218010.1109/TAC.2008.9301992459593ShenB.WangZ.HungY. S.Distributed H∞-consensus filtering in sensor networks with multiple missing measurements: the finite-horizon case201046101682168810.1016/j.automatica.2010.06.0252877323ShenB.WangZ.ShuH.WeiG.Robust H∞ finite-horizon filtering with randomly occurred nonlinearities and quantization effects201046111743175110.1016/j.automatica.2010.06.0412878188ZBL1218.93103XuY.ZhouW.FangJ.LuH.Structure identification and adaptive synchronization of uncertain general complex dynamical networks200937422722782-s2.0-7184910332410.1016/j.physleta.2009.10.079XuY.ZhouW.FangJ.SunW.Adaptive synchronization of the complex dynamical network with non-derivative and derivative coupling201037415-16167316772-s2.0-7794929210110.1016/j.physleta.2010.02.002TangY.QiuR.FangJ.MiaoQ.XiaM.Adaptive lag synchronization in unknown stochastic chaotic neural networks with discrete and distributed time-varying delays2008372244425443310.1016/j.physleta.2008.04.0322432547ZBL1221.82078ZhouJ.XiangL.LiuZ.Synchronization in complex delayed dynamical networks with impulsive effects200738426846922-s2.0-3454805110810.1016/j.physa.2007.05.060WuX.Synchronization-based topology identification of weighted general complex dynamical networks with time-varying coupling delay2008387499710082-s2.0-3664901709710.1016/j.physa.2007.10.030WuZ.-Y.LiK.-Z.FuX.-C.Parameter identification of dynamical networks with community structure and multiple coupling delays201015113587359210.1016/j.cnsns.2009.12.0282646188ZBL1222.93058TangY.FangJ.XiaM.GuX.Synchronization of Takagi-Sugeno fuzzy stochastic discrete-time complex networks with mixed time-varying delays201034484385510.1016/j.apm.2009.07.0152579599ZBL1185.93145XuY.ZhouW.FangJ.Topology identification of the modified complex dynamical network with non-delayed and delayed coupling2012681-219520510.1007/s11071-011-0217-xPengH.WeiN.LiL.XieW.YangY.Models and synchronization of time-delayed complex dynamical networks with multi-links based on adaptive control201037423233523392-s2.0-7795092437610.1016/j.physleta.2010.03.052ZhouJ.XiangL.LiuZ.Synchronization in complex delayed dynamical networks with impulsive effects200738426846922-s2.0-3454805110810.1016/j.physa.2007.05.060LiK.LaiC. H.Adaptive-impulsive synchronization of uncertain complex dynamical networks2008372101601160610.1016/j.physleta.2007.10.0202396238ZBL1217.05210CaiS.ZhouJ.XiangL.LiuZ.Robust impulsive synchronization of complex delayed dynamical networks200837230499049952-s2.0-4644909419410.1016/j.physleta.2008.05.077YangT.2001Berlin, GermanySpringer1850661LüJ.ChenG.A time-varying complex dynamical network model and its controlled synchronization criteria200550684184610.1109/TAC.2005.8492332142000ZhouJ.LüJ.LuJ.Adaptive synchronization of an uncertain complex dynamical network200651465265610.1109/TAC.2006.8727602228029LüJ.YuX.ChenG.Chaos synchronization of general complex dynamical networks20043341-228130210.1016/j.physa.2003.10.0522044940ZhouJ.LüJ.LuJ.Pinning adaptive synchronization of a general complex dynamical network2008444996100310.1016/j.automatica.2007.08.0162530942LüJ.ChenG.A new chaotic attractor coined200212365966110.1142/S02181274020046201894886ZBL1063.34510XuY.LiB.WangY.ZhouW.FangJ.A new four-scroll chaotic attractor consisted of two-scroll transient chaotic and two-scroll ultimate chaotic201220121243832810.1155/2012/438328XuY.ZhouW.FangJ.MaJ.WangY.Generating a new chaotic attractor by feedback controlling method201134172159216610.1002/mma.15132861759ZBL1230.37111LuW.ChenT.ChenG.Synchronization analysis of linearly coupled systems described by differential equations with a coupling delay2006221211813410.1016/j.physd.2006.07.0202265856ZBL1111.34056