Cluster lag synchronization of delayed heterogeneous complex dynamical networks via intermittent pinning control

Fan Yang; Huaqing Li; Guo Chen; Dawen Xia; Qi Han

doi:10.1007/s00521-018-3618-7

Neural Computing and Applications

November 2019, Volume 31, Issue 11, pp 7945–7961 | Cite as

Cluster lag synchronization of delayed heterogeneous complex dynamical networks via intermittent pinning control

Authors
Authors and affiliations

Fan Yang
Huaqing Li
Guo Chen
Dawen Xia
Qi Han

Original Article

First Online: 17 July 2018

247 Downloads

Abstract

This paper investigates the problem of cluster lag synchronization in the heterogeneous dynamical networks by using an intermittent pinning control strategy. Previous related works mainly focused on the time-varying delays in the self-dynamics, which was not consistent with the real world. The transmission delay in the communication channels is considered in this paper. We present several criteria to guarantee cluster lag synchronization without assuming the coupling matrix being symmetric and irreducible. A decentralized adaptive intermittent pinning control scheme is employed to reduce the control cost. An effective pinned-cluster selection scheme is adopted to guide what kind of clusters should be pinned preferentially. Two simulations are proposed to verify the correctness of the theoretical results.

Keywords

Lag synchronization Cluster synchronization Adaptive control Intermittent pinning control Heterogeneous complex networks

This is a preview of subscription content, to check access.

Notes

Acknowledgements

The work described in this paper was supported in part by the Special Financial Support from China Postdoctoral Science Foundation under Grant 2017T100670, in part by the China Postdoctoral Science Foundation under Grant 2016M590852, in part by the Special Financial Support from Chongqing Postdoctoral Science Foundation under Grant Xm2017100, in part by the National Natural Science Foundation of China under Grants 61773321, 61762020 and 61503050, in part by the Science and Technology Foundation of Guizhou under Grants. QKHJC20161076 and QKHJC20181083, in part by the Science and Technology Top-notch Talents Support Project of Colleges and Universities in Guizhou under Grant QJHKY2016065 and in part by the High-level Innovative Talents Project of Guizhou under Grant QRLF201621.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Appendix 1

Proof of Theorem 1

Let $E\left( t \right) = \left( {E_{1}^{\rm T} \left( t \right),E_{2}^{\rm T} \left( t \right), \ldots ,E_{m}^{\rm T} \left( t \right)} \right)^{\rm T}$, where $E_{1} \left( t \right) = \left( {e_{1,1}^{\rm T} \left( t \right), \ldots ,e_{{r_{1} ,1}}^{\rm T} \left( t \right)} \right)^{\rm T}$, $E_{2} \left( t \right) = \left( {e_{{r_{1} + 1,2}}^{\rm T} \left( t \right), \ldots ,e_{{r_{1} + r_{2} ,2}}^{\rm T} \left( t \right)} \right)^{\rm T} , \ldots .E_{m} \left( t \right) = \left( {e_{{r_{1} + r_{2} + \cdots + r_{m - 1} + 1,m}}^{\rm T} \left( t \right), \ldots ,e_{N,m}^{\rm T} \left( t \right)} \right)^{\rm T} .$

Construct a Lyapunov candidate as follows, where $\otimes$ is defined as Kronecker product.

$$\begin{aligned} W\left( t \right) & = \frac{1}{2}E^{\rm T} \left( t \right)\left( {I_{N} \otimes I_{n} } \right)E\left( t \right) = \frac{1}{2}\sum\nolimits_{k = 1}^{m} {E_{k}^{\rm T} \left( t \right)\left( {I_{{r_{k} }} \otimes I_{n} } \right)E_{k} \left( t \right)} \\ & = \frac{1}{2}\sum\nolimits_{k = 1}^{m} {\sum\nolimits_{{i \in C_{k} }} {e_{ik}^{\rm T} \left( t \right)e_{ik} \left( t \right)} } \\ \end{aligned}$$

When $t \in \left[ {\omega T,\left( {\omega + \theta } \right)T} \right)$, $\omega = 0,1,2, \ldots$, the time derivative of $W\left( t \right)$ along the trajectory of (9) at time $t$ can be given as

$$\begin{aligned} \dot{W}\left( t \right) & = \sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {e_{ik}^{\text{T}} \left( t \right)\tilde{f}_{k}^{{\tau_{k} ,\sigma }} \left( {t,x_{i} ,s_{k} } \right)} } + c\sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {\sum\limits_{p = 1}^{m} {\sum\limits_{{j \in C_{p} }} {b_{ij} e_{ik}^{\text{T}} \left( t \right)\varGamma e_{jp} \left( t \right)} } } } \\ & \quad - c\sum\limits_{k = 1}^{l} {E_{k}^{\text{T}} \left( t \right)\left( {d_{k} I_{{r_{k} }} \otimes \varGamma } \right)E_{k} \left( t \right)} . \\ \end{aligned}$$

(28)

In virtue of Assumption 2, we can get

$$\begin{aligned} & \sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {e_{ik}^{\text{T}} \left( t \right)\tilde{f}_{k}^{{\tau_{k} ,\sigma }} \left( {t,x_{i} ,s_{k} } \right)} } \\ & \quad \le \sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {L_{k}^{0} e_{ik}^{\text{T}} \left( t \right)\varGamma e_{ik} \left( t \right)} } + \sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {L_{k}^{\tau } e_{ik}^{\text{T}} \left( {t - \tau_{k} \left( t \right)} \right)\varGamma e_{ik} \left( {t - \tau_{k} \left( t \right)} \right)} } \\ & \quad = \sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( t \right)\left( {L_{k}^{0} I_{{r_{k} }} \otimes \varGamma } \right)E_{k} \left( t \right)} + \sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( {t - \tau_{k} \left( t \right)} \right)\left( {L_{k}^{\tau } I_{{r_{k} }} \otimes \varGamma } \right)E_{k} \left( {t - \tau_{k} \left( t \right)} \right)} . \\ \end{aligned}$$

(29)

Noticing that $\sum\nolimits_{p = 1}^{m} {\sum\nolimits_{{i = C_{p} }} {b_{ij} } } = \sum\nolimits_{j = 1}^{N} {b_{ij} } = 0$, $i = 1,2, \ldots ,N$ with $\tilde{\varGamma } = \text{diag} \left( {\sqrt {\gamma_{1} } ,\sqrt {\gamma_{2} } , \ldots ,\sqrt {\gamma_{n} } } \right)$, one has

$$\begin{aligned} & \sum\limits_{k = 1}^{m} {\sum\limits_{{i = C_{k} }} {\sum\limits_{p = 1}^{m} {\sum\limits_{{j \in C_{p} }} {b_{ij} e_{ik}^{\text{T}} \left( t \right)\varGamma e_{jp} \left( t \right)} } } } = \sum\limits_{k = 1}^{m} {\sum\limits_{p = 1}^{m} {E_{k}^{\text{T}} \left( t \right)\left( {B_{kp} \otimes \varGamma } \right)E_{p} \left( t \right)} } \\ & \quad = \sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( t \right)\left( {B_{kk}^{s} \otimes \varGamma } \right)E_{k} \left( t \right)} + \sum\limits_{k = 1}^{m} {\sum\limits_{p = 1,p \ne k}^{m} {\sum\limits_{{i = C_{k} }} {\sum\limits_{{j \in C_{p} }} {b_{ij} \left( {\tilde{\varGamma }e_{ik} \left( t \right)} \right)^{\text{T}} \left( {\tilde{\varGamma }e_{jp} \left( t \right)} \right)} } } } \\ & \quad \le \sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( t \right)\left( {B_{kk}^{s} \otimes \varGamma } \right)E_{k} \left( t \right)} + \frac{1}{2}\sum\limits_{k = 1}^{m} {\sum\limits_{p = 1,p \ne k}^{m} {\sum\limits_{{i = C_{k} }} {\sum\limits_{{j \in C_{p} }} {b_{ij} \left( {e_{ik}^{\text{T}} \left( t \right)\varGamma e_{ik} \left( t \right) + e_{jp}^{\text{T}} \left( t \right)\varGamma e_{jp} \left( t \right)} \right)} } } } \\ & \quad = \sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( t \right)\left( {B_{kk}^{s} \otimes \varGamma } \right)E_{k} \left( t \right)} - \frac{1}{2}\sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( t \right)\left( {A_{k} \otimes \varGamma } \right)E_{k} \left( t \right)} + \frac{1}{2}\sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( t \right)\left( {\varPsi_{k} \otimes \varGamma } \right)E_{k} \left( t \right)} \\ & \quad = \sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( t \right)\left( {\left( {B_{kk}^{s} + \frac{1}{2}\varPsi_{k} - \frac{1}{2}A_{k} } \right) \otimes \varGamma } \right)E_{k} \left( t \right)} \\ \end{aligned}$$

(30)

Substituting (29) and (30) into (28), we get

$$\dot{W}\left( t \right) \le cE^{\rm T} \left( t \right)\left( {\left( {Z - D} \right) \otimes \varGamma } \right)E\left( t \right) + \frac{q}{2}\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} \sum\limits_{k = 1}^{m} {E_{k}^{\text{T}} \left( s \right)E_{k} \left( s \right)} } \right).$$

(31)

According to condition $\left( \text{i} \right)$ and the properties of the Kronecker product of matrix [42], from (31), one has

$$\begin{aligned} \dot{W}\left( t \right) & \le E^{\rm T} \left( t \right)\left( {\left( {cZ - cD + a_{1} I_{N} } \right) \otimes \varGamma } \right)E\left( t \right) - a_{1} E^{\rm T} \left( t \right)\left( {I_{N} \otimes \varGamma } \right)E\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right) \\ & \le - 2a_{1} \lambda_{\hbox{min} } \left( \varGamma \right)W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right) = - p_{1} W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right). \\ \end{aligned}$$

Similarly, when $t \in \left[ {\left( {\omega + \theta } \right)T,\left( {\omega + 1} \right)T} \right)$, $\omega = 0,1,2, \ldots$, using condition $\left( {\text{ii} } \right)$, we can deduce

$$\begin{aligned} \dot{W}\left( t \right) & \le E^{\rm T} \left( t \right)\left( {\left( {cZ - a_{2} I_{N} } \right) \otimes \varGamma } \right)E\left( t \right) + a_{2} E^{\rm T} \left( t \right)\left( {I_{N} \otimes \varGamma } \right)E\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right) \\ & \le 2a_{2} \lambda_{\hbox{max} } \left( \varGamma \right)W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right) = p_{2} W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right). \\ \end{aligned}$$

As a result, we have

$$\left\{ {\begin{array}{ll} {\dot{W}\left( t \right) \le - p_{1} W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right),} & {\omega T \le t < \left( {\omega + \theta } \right)T,} \\ {\dot{W}\left( t \right) \le p_{2} W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right),} & {\left( {\omega + \theta } \right)T \le t < \left( {\omega + 1} \right)T.} \\ \end{array} } \right.$$

(32)

Next, we will prove that conditions (iii)–(iv) imply that, for $t \ge 0$, we have $\dot{W}\left( t \right) \le \left( {\mathop {\sup }\nolimits_{ - \tau \le s \le 0} W\left( s \right)} \right)e^{ - \varpi t}$.

Define $\zeta \left( \lambda \right) = \lambda - p_{1} + qe^{\lambda \tau }$, where $p_{1} > q > 0$ such that $\zeta \left( 0 \right) < 0$, $\zeta \left( { + \infty } \right) > 0$, $\zeta^{\prime}\left( \lambda \right) > 0$. Utilizing the monotonicity and continuity of $\zeta \left( \lambda \right)$, the equation $\lambda - p_{1} + qe^{\lambda \tau } = 0$ has a unique positive solution $\lambda > 0$. Let $M_{0} = \mathop {\sup }\nolimits_{ - \tau \le s \le 0} W\left( s \right)$, $V\left( t \right) = e^{\lambda t} W\left( t \right)$, $t \ge 0$. Let $S\left( t \right) = V\left( t \right) - hM_{0}$, where $h > 1$ is a constant. Apparently, for any $t \in \left[ { - \tau ,0} \right]$, one has

$$S\left( t \right) < 0.$$

(33)

Then, we indicate for all $t \in \left[ {0,\theta T} \right]$, that

$$S\left( t \right) < 0.$$

(34)

From (33), there exist a $t_{0} \in \left[ {0,\theta T} \right]$ such that

$$S\left( {t_{0} } \right) = 0,{\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} \dot{S}\left( {t_{0} } \right) \ge 0,$$

(35)

$$S\left( t \right) < 0,{\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} - \tau \le t < t_{0} .$$

(36)

By (32), (35) and (36), we have

$$\dot{S}\left( {t_{0} } \right) = \lambda V\left( {t_{0} } \right) + e^{{\lambda t_{0} }} \dot{W}\left( {t_{0} } \right) \le \left( {\lambda - p_{1} } \right)V\left( {t_{0} } \right) + qe^{{\lambda t_{0} }} \left( {\mathop {\sup }\limits_{{t_{0} - \tau \le s \le t_{0} }} W\left( s \right)} \right).$$

(37)

On the other hand, from (35) and (36), we have $V\left( t \right) < hM_{0}$, $- \tau \le t < t_{0}$ and $V\left( {t_{0} } \right) = hM_{0}$. As a result, we get $W\left( t \right) < hM_{0} e^{ - \lambda t}$, $- \tau \le t < t_{0}$, it leads to $\left( {\sup_{{t_{0} - \tau \le s \le t_{0} }} W\left( s \right)} \right) < e^{\lambda \tau } hM_{0} e^{{ - \lambda t_{0} }}$. This means $e^{{\lambda t_{0} }} \left( {\sup_{{t_{0} - \tau \le s \le t_{0} }} W\left( s \right)} \right) < e^{\lambda \tau } hM_{0} = e^{\lambda \tau } V\left( {t_{0} } \right)$.

From (37), it can be deduced that $\dot{S}\left( {t_{0} } \right) < \left( {\lambda - p_{1} + qe^{\lambda \tau } } \right)V\left( {t_{0} } \right) = 0$. This is a contradiction with inequality (35), thus, (34) is established. With (33), we get that

$$W\left( t \right) < hM_{0} e^{ - \lambda t} ,\quad \forall t \in \left[ { - \tau ,\theta T} \right]$$

(38)

Let $\vartheta = p_{1} + p_{2}$, for $t \in \left[ {\theta T,T} \right)$, we have $Q\left( t \right) = V\left( t \right) - hM_{0} e^{{\vartheta \left( {t - \theta T} \right)}} < 0$. Otherwise, there exist a $t_{1} \in \left[ {\theta T,T} \right)$, such that

$$Q\left( {t_{1} } \right) = 0,\quad \dot{Q}\left( {t_{1} } \right) \ge 0,$$

(39)

$$Q\left( t \right) < 0,\quad \theta T \le t < t_{1} .$$

(40)

For $\tau > 0$, according to (38)–(40), we always have

$$\mathop {\sup }\limits_{{t_{1} - \tau \le s \le t_{1} }} W\left( s \right) < e^{\lambda \tau } W\left( {t_{1} } \right).$$

Then, one has $\dot{Q}\left( {t_{1} } \right) < 0$, which is contradicts with (39). Thus, for $t \in \left[ {\theta T,T} \right)$,

$$V\left( t \right) < hM_{0} e^{{\vartheta \left( {1 - \theta } \right)T}} .$$

With (33) and (34), we have

$$V\left( t \right) < hM_{0} e^{{\vartheta \left( {1 - \theta } \right)T}} ,\quad {\text{for}}\quad t \in \left[ { - \tau ,T} \right).$$

Similarly, we can prove

$$V\left( t \right) < hM_{0} e^{{\vartheta \left( {1 - \theta } \right)T}} ,\;\;{\text{for}}\;\;t \in \left[ {T,\left( {1 + \theta } \right)T} \right)\;\;{\text{and}}\;\;V\left( t \right) < hM_{0} e^{{\vartheta \left( {t - 2\theta } \right)T}} ,\;\;{\text{for}}\;\;t \in \left[ {\left( {1 + \theta } \right)T,2T} \right).$$

Through a mathematical induction, for any integer $\omega = 0,1,2, \ldots$, we can derive the estimates of $V\left( t \right)$ as

$$\left\{ {\begin{array}{ll} {V\left( t \right) < hM_{0} e^{{\omega \vartheta \left( {1 - \theta } \right)T}} \le hM_{0} e^{{\vartheta \left( {1 - \theta } \right)t}} ,} & {\omega T \le t < \left( {\omega + \theta } \right)T,} \\ {V\left( t \right) < hM_{0} e^{{\vartheta \left[ {t - \left( {m + 1} \right)\theta T} \right]}} \le hM_{0} e^{{\vartheta \left( {1 - \theta } \right)t}} ,} & {\left( {\omega + \theta } \right)T \le t < \left( {\omega + 1} \right)T.} \\ \end{array} } \right.$$

Let $h \to 1$, from the definition of $V\left( t \right)$, we have

$$W\left( t \right) \le M_{0} e^{{ - \left[ {\lambda - \vartheta \left( {1 - \theta } \right)} \right]t}} = \left( {\mathop {\sup }\limits_{ - \tau \le s \le 0} W\left( s \right)} \right)e^{ - \varpi t} ,\quad t \ge 0.$$

Therefore, for the error dynamical system (9) the zero solution is globally exponentially stable. The proof of Theorem 1 is thus completed.

Appendix 2

Proof of Theorem 3

Construct a piecewise Lyapunov candidate function

$$W\left( t \right) = \frac{1}{2}\sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {e_{ik}^{\text{T}} \left( t \right)e_{ik} \left( t \right)} } + \frac{1}{2}\varPi \left( t \right)\sum\limits_{k = 1}^{l} {ce^{{ - p_{1} t}} \frac{{\left( {d_{k} \left( t \right) - d_{k}^{*} } \right)^{2} }}{{h_{k} }}} ,$$

where $d_{k}^{ * } > 0$, $k = 1,2, \ldots ,l$ are constants, and $\varPi \left( \cdot \right)$ is a piecewise function defined as

$$\varPi \left( s \right) = \left\{ {\begin{array}{ll} {1,} & { - \tau \le s \le 0,} \\ {e^{{p_{1} \omega T}} ,} & {\omega T \le s < \left( {\omega + 1} \right)T,\omega = 0,1,2, \ldots } \\ \end{array} } \right.$$

Obviously, while $\omega = 0,1,2, \ldots$, we have

$$W\left( t \right) = \left\{ {\begin{array}{ll} {\frac{1}{2}\sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {e_{ik}^{\text{T}} \left( t \right)e_{ik} \left( t \right)} } + \frac{1}{2}e^{{ - p_{1} \left( {t - \omega T} \right)}} \sum\limits_{k = 1}^{l} {\frac{c}{{h_{k} }}\left( {d_{k} \left( t \right) - d_{k}^{*} } \right)^{2} } ,} & {\omega T \le t < \left( {\omega + \theta } \right)T,} \\ {\frac{1}{2}\sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {e_{ik}^{\text{T}} \left( t \right)e_{ik} \left( t \right)} } + \frac{1}{2}e^{{ - p_{1} \left( {t - \omega T} \right)}} \sum\limits_{k = 1}^{l} {\frac{c}{{h_{k} }}d_{k}^{{*^{2} }} } ,} & {\left( {\omega + \theta } \right)T \le t < \left( {\omega + 1} \right)T.} \\ \end{array} } \right.$$

With Assumptions 1 and 2, the time derivative of $W\left( t \right)$ along the trajectory of (9) can be calculated as follows. When $t \in \left[ {\omega T,\left( {\omega + \theta } \right)T} \right)$, $\omega = 0,1,2, \ldots$, we have

$$\begin{aligned} \dot{W}\left( t \right) & \le E^{\rm T} \left( t \right)\left( {\left( {cZ - cD^{ * } + a_{1}^{0} I_{N} } \right) \otimes \varGamma } \right)E\left( t \right) + \frac{q}{2}\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} \sum\nolimits_{k = 1}^{m} {E_{k}^{\rm T} \left( s \right)E_{k} \left( s \right)} } \right) \\ & \quad - a_{1}^{0} E^{\rm T} \left( t \right)\left( {I_{N} \otimes \varGamma } \right)E\left( t \right) - \frac{{p_{1} }}{2}e^{{ - p_{1} \left( {t - \omega T} \right)}} \sum\nolimits_{k = 1}^{l} {\frac{c}{{h_{k} }}\left( {d_{k} \left( t \right) - d_{k}^{ * } } \right)^{2} } , \\ \end{aligned}$$

(41)

where $D^{ * }$ is a modified block diagonal matrix of $D$ through replacing the principal-diagonal sub-matrices $d_{k} I_{{r_{k} }}$ by $\left( {e^{{ - p_{1} \theta T}} d_{k}^{ * } } \right)I_{{r_{k} }} ,k = 1,2, \ldots ,l$.

Set $Q^{ * } = a_{1}^{0} I_{N} + cZ$, $a_{1}^{0} I_{N} + cZ - cD^{ * } = Q^{ * } - cD^{ * } = \left( {\begin{array}{*{20}c} {G^{ * } - c\tilde{D}^{ * } } & 0 \\ {0^{\rm T} } & {Q_{r}^{ * } } \\ \end{array} } \right)$, where $G^{ * } = \left( {\begin{array}{llll} {cZ_{11} + a_{1}^{0} I_{{r_{1} }} } & 0 & \cdots & 0 \\ 0 & {cZ_{22} + a_{1}^{0} I_{{r_{2} }} } & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \cdots & {cZ_{ll} + a_{1}^{0} I_{{r_{l} }} } \\ \end{array} } \right)$, $\tilde{D}^{ * } = e^{{ - p_{1} \theta T}} \left( {\begin{array}{llll} {d_{1}^{ * } I_{{r_{1} }} } & 0 & \cdots & 0 \\ 0 & {d_{2}^{ * } I_{{r_{2} }} } & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \cdots & {d_{l}^{ * } I_{{r_{l} }} } \\ \end{array} } \right)$, and $Q_{r}^{ * }$ is the minor matrix of $Q^{ * }$ by removing its first $\left( {r_{1} + r_{2} + \cdots + r_{l} } \right)$ row–column pairs. It is obviously that $Q^{ * }$ is a real symmetric matrix. From the pinning condition $\left( {{\text{i}}^{\prime \prime } } \right)$, we conclude that

$$\begin{aligned} \lambda_{\hbox{max} } \left( {cZ_{kk} + a_{1}^{0} I_{{r_{k} }} } \right) & = \lambda_{\hbox{max} } \left( {c\varTheta_{kk} + L_{k}^{0} I_{{r_{k} }} + a_{1}^{0} I_{{r_{k} }} } \right) \\ & < c\lambda_{\hbox{max} } \left( {\varTheta_{kk} } \right) + L_{k}^{0} + a_{1}^{0} < 0,\quad l + 1 \le k \le m. \\ \end{aligned}$$

This means $Q_{r}^{ * } < 0$. As a result, for $k = 1,2, \cdots ,l$, when $d_{k}^{ * } > 0$ are large enough such that $d_{k}^{ * } > \frac{{e^{{p_{1} \theta T}} \lambda_{\hbox{max} } \left( {G^{ * } } \right)}}{c}$. It is easy to get $Q^{ * } - cD^{ * } < 0$. This follows Lemma 2 directly.

With (41), we can deduce that

$$\begin{aligned} \dot{W}\left( t \right) & \le - \frac{{p_{1} }}{2}\sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {e_{ik}^{\text{T}} \left( t \right)e_{ik} \left( t \right)} } - \frac{{p_{1} }}{2}e^{{ - p_{1} \left( {t - \omega T} \right)}} \sum\limits_{k = 1}^{l} {\frac{c}{{h_{k} }}\left( {d_{k} \left( t \right) - d_{k}^{*} } \right)^{2} } \\ & \quad + \frac{q}{2}\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} \sum\limits_{k = 1}^{m} {\sum\limits_{{i \in C_{k} }} {e_{ik}^{\text{T}} \left( s \right)e_{ik} \left( s \right)} } } \right) \\ & \le - p_{1} W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right). \\ \end{aligned}$$

In the same way, when $t \in \left[ {\left( {\omega + \theta } \right)T,\left( {\omega + 1} \right)T} \right)$, $\omega = 0,1,2, \ldots$, we obtain that

$$\dot{W}\left( t \right) \le p_{2} W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right).$$

Namely,

$$\left\{ {\begin{array}{ll} {\dot{W}\left( t \right) \le - p_{1} W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right),} & {\omega T \le t < \left( {\omega + \theta } \right)T,} \\ {\dot{W}\left( t \right) \le p_{2} W\left( t \right) + q\left( {\mathop {\sup }\limits_{t - \tau \le s \le t} W\left( s \right)} \right),} & {\left( {\omega + \theta } \right)T \le t < \left( {\omega + 1} \right)T.} \\ \end{array} } \right.$$

Then, by a similar proof of Theorem 1, we can prove that condition (15) means

$$\dot{W}\left( t \right) \le \left( {\mathop {\sup }\limits_{ - \tau \le s \le 0} W\left( s \right)} \right)e^{ - \varpi t} ,\quad t \ge 0.$$

Therefore, the global cluster lag synchronization of the delayed dynamical network (6) under the adaptive intermittent pinned controllers (16)–(17) is realizable. The proof of Theorem 3 is thus completed.

References

1.
Strogatz SH (2001) Exploring complex networks. Nature 410(6825):268–276zbMATHCrossRefGoogle Scholar
2.
Newman MEJ (2003) The structure and function of complex networks. SIAM Rev 45(2):167–256MathSciNetzbMATHCrossRefGoogle Scholar
3.
Wang J, Wu H (2012) Local and global exponential output synchronization of complex delayed dynamical networks. Nonlinear Dyn 67(1):497–504MathSciNetzbMATHCrossRefGoogle Scholar
4.
Ning D, Wu X, Lu J et al (2015) Driving-based generalized synchronization in two-layer networks via pinning control. Chaos 25(11):113104MathSciNetzbMATHCrossRefGoogle Scholar
5.
Li H, Liao X, Chen G et al (2015) Event-triggered asynchronous intermittent communication strategy for synchronization in complex dynamical networks. Neural Netw 66(C):1–10zbMATHGoogle Scholar
6.
Li H, Liao X, Huang T et al (2015) Second-order global consensus in multiagent networks with random directional link failure. IEEE Trans Neural Netw Learn Syst 26(3):565–575MathSciNetCrossRefGoogle Scholar
7.
Cai G, Jiang S, Cai S et al (2015) Cluster synchronization of overlapping uncertain complex networks with time-varying impulse disturbances. Nonlinear Dyn 80(1–2):503–513MathSciNetzbMATHCrossRefGoogle Scholar
8.
Yang S, Li C, Huang T (2017) Synchronization of coupled memristive chaotic circuits via state-dependent impulsive control. Nonlinear Dyn 88(1):115–129zbMATHCrossRefGoogle Scholar
9.
Cheng J, Park J et al (2018) An asynchronous operation approach to event-triggered control for fuzzy markovian jump systems with general switching policies. IEEE Trans Fuzzy Syst 26(1):6–18CrossRefGoogle Scholar
10.
Lai H, Huang Y (2015) Chaotic secure communication based on synchronization control of chaotic pilot signal. Computational intelligence and intelligent systems. Springer, SingaporeGoogle Scholar
11.
Prakash M, Balasubramaniam P, Lakshmanan S (2016) Synchronization of markovian jumping inertial neural networks and its applications in image encryption. Neural Netw 83:86–93CrossRefGoogle Scholar
12.
Aghababa MP, Aghababa HP (2013) Robust synchronization of a chaotic mechanical system with nonlinearities in control inputs. Nonlinear Dyn 73(1–2):363–376MathSciNetzbMATHCrossRefGoogle Scholar
13.
Zhang J, Ma Z, Zhang G (2013) Cluster synchronization induced by one-node clusters in networks with asymmetric negative couplings. Chaos 23(4):043128MathSciNetzbMATHCrossRefGoogle Scholar
14.
Zhou L, Wang C, Du S et al (2017) Cluster synchronization on multiple nonlinearly coupled dynamical subnetworks of complex networks with nonidentical nodes. IEEE Trans Neural Netw Learn Syst 28(3):1–14MathSciNetCrossRefGoogle Scholar
15.
Li H, Liao X, Chen G et al (2017) Attraction region seeking for power grids. IEEE Trans Circuits Syst II Express Briefs 64(2):201–205CrossRefGoogle Scholar
16.
Huang T, Li C, Duan S et al (2012) Robust exponential stability of uncertain delayed neural networks with stochastic perturbation and impulse effects. IEEE Trans Neural Netw Learn Syst 23(6):866–875CrossRefGoogle Scholar
17.
Zeng Z, Huang T, Zheng W (2010) Multistability of recurrent neural networks with time-varying delays and the piecewise linear activation function. IEEE Trans Neural Netw Learn Syst 21(8):1371–1377CrossRefGoogle Scholar
18.
Huang T, Li C, Yu W et al (2009) Synchronization of delayed chaotic systems with parameter mismatches by using intermittent linear state feedback. Nonlinearity 22(3):569–584MathSciNetzbMATHCrossRefGoogle Scholar
19.
Chen T, Liu X, Lu W (2007) Pinning complex networks by a single controller. IEEE Trans Circuits Syst I Regular Papers 54(6):1317–1326MathSciNetzbMATHCrossRefGoogle Scholar
20.
Yu W, Chen G, Lü J et al (2013) Synchronization via pinning control on general complex networks. SIAM J Control Optim 51(2):1395–1416MathSciNetzbMATHCrossRefGoogle Scholar
21.
Shi L, Zhu H, Zhong S et al (2017) Cluster synchronization of linearly coupled complex networks via linear and adaptive feedback pinning controls. Nonlinear Dyn 88(2):859–870zbMATHCrossRefGoogle Scholar
22.
Liu X, Chen T (2015) Synchronization of linearly coupled networks with delays via aperiodically intermittent pinning control. IEEE Trans Neural Netw Learn Syst 26(10):2396–2407MathSciNetCrossRefGoogle Scholar
23.
Ali MS, Yogambigai J (2016) Exponential stability of semi-markovian switching complex dynamical networks with mixed time varying delays and impulse control. Neural Process Lett 46:1–21Google Scholar
24.
Wang X, She K, Zhong S et al (2016) New result on synchronization of complex dynamical networks with time-varying coupling delay and sampled-data control. Neurocomputing 214:508–515CrossRefGoogle Scholar
25.
Hu A, Cao J, Hu M et al (2015) Cluster synchronization of complex networks via event-triggered strategy under stochastic sampling. Phys A 434(15):99–110MathSciNetzbMATHCrossRefGoogle Scholar
26.
Liu X, Chen T (2011) Cluster synchronization in directed networks via intermittent pinning control. IEEE Trans Neural Netw 22(7):1009–1020CrossRefGoogle Scholar
27.
Mei J, Jiang M, Wu Z et al (2015) Periodically intermittent controlling for finite-time synchronization of complex dynamical networks. Nonlinear Dyn 79(1):295–305zbMATHCrossRefGoogle Scholar
28.
Cai S, Zhou P, Liu Z (2015) Intermittent pinning control for cluster synchronization of delayed heterogeneous dynamical networks. Nonlinear Anal Hybrid Syst 18:134–155MathSciNetzbMATHCrossRefGoogle Scholar
29.
He W, Zhang S, Ge S (2014) Adaptive control of a flexible crane system with the boundary output constraint. IEEE Trans Ind Electron 61(8):4126–4133CrossRefGoogle Scholar
30.
He W, Ge S (2015) Vibration control of a flexible beam with output constraint. IEEE Trans Ind Electron 62(8):5023–5030CrossRefGoogle Scholar
31.
Zhao Z, He W, Yin Z et al (2017) Spatial trajectory tracking control of a fully actuated helicopter in known static environment. J Intell Robot Syst 85(1):1–18CrossRefGoogle Scholar
32.
He W, Zhang S, Ge S (2014) Robust adaptive control of a thruster assisted position mooring system. Automatica 50(7):1843–1851MathSciNetzbMATHCrossRefGoogle Scholar
33.
He W, Chen Y, Yin Z (2016) Adaptive neural network control of an uncertain robot with full-state constraints. IEEE Trans Cybern 46(3):620–629CrossRefGoogle Scholar
34.
Zhang X, Ma H, Yang C (2017) Decentralised adaptive control of a class of hidden leader–follower non-linearly parameterised coupled mass. IET Control Theory A 11(17):3016–3025MathSciNetCrossRefGoogle Scholar
35.
Zhou P, Cai S (2017) Pinning synchronization of complex directed dynamical networks under decentralized adaptive strategy for aperiodically intermittent control. Nonlinear Dyn 90(1):287–299MathSciNetzbMATHCrossRefGoogle Scholar
36.
Jiang S, Lu X (2016) Synchronization analysis of coloured delayed networks under decentralized pinning intermittent control. Pramana 86(6):1243–1251CrossRefGoogle Scholar
37.
Wang K, Fu X, Li K (2009) Cluster synchronization in community networks with nonidentical nodes. Chaos 19(2):023106MathSciNetzbMATHCrossRefGoogle Scholar
38.
Hu C, Jiang H (2012) Cluster synchronization for directed community networks via pinning partial schemes. Chaos Soliton Fractals 45(11):1368–1377MathSciNetzbMATHCrossRefGoogle Scholar
39.
Su H, Rong Z, Chen MZQ et al (2013) Decentralized adaptive pinning control for cluster synchronization of complex dynamical networks. IEEE Trans Cybern 43(1):394–399CrossRefGoogle Scholar
40.
Wang Y, Cao J (2013) Cluster synchronization in nonlinearly coupled delayed networks of non-identical dynamic systems. Nonlinear Anal Real 14(1):842–851MathSciNetzbMATHCrossRefGoogle Scholar
41.
Xia W, Cao J (2009) Pinning synchronization of delayed dynamical networks via periodically intermittent control. Chaos 19(1):013120MathSciNetzbMATHCrossRefGoogle Scholar
42.
Cai S, Zhou J, Xiang L et al (2008) Robust impulsive synchronization of complex delayed dynamical networks. Phys Lett A 372(30):4990–4995zbMATHCrossRefGoogle Scholar
43.
Horn RA, Johnson CR (1985) Matrix analysis. Cambridge University Press, CambridgezbMATHCrossRefGoogle Scholar
44.
Song Q, Cao J (2010) On pinning synchronization of directed and undirected complex dynamical networks. IEEE Trans Circuits Syst I Regular Papers 57(3):672–680MathSciNetCrossRefGoogle Scholar

Copyright information

Authors and Affiliations

Fan Yang
- 1
Huaqing Li
- 1
Email author
Guo Chen
- 2
Dawen Xia
- 3
Qi Han
- 4

1.Chongqing Key Laboratory of Nonlinear Circuits and Intelligent Information Processing, College of Electronic and Information EngineeringSouthwest UniversityChongqingPeople’s Republic of China
2.School of Electrical Engineering and Computer ScienceUniversity of NewcastleSydneyAustralia
3.College of Data Science and Information EngineeringGuizhou Minzu UniversityGuiyangPeople’s Republic of China
4.College of Electrical and Information EngineeringChongqing University of Science and TechnologyChongqingPeople’s Republic of China

Cluster lag synchronization of delayed heterogeneous complex dynamical networks via intermittent pinning control

Abstract

Keywords

Notes

Acknowledgements

Compliance with ethical standards

Conflict of interest

References

Copyright information

Authors and Affiliations

Personalised recommendations

Cite article