Loading [MathJax]/jax/output/SVG/jax.js
Research article Special Issues

Average-delay impulsive control for synchronization of uncertain chaotic neural networks with variable delay impulses


  • This paper investigated the synchronization issue of uncertain chaotic neural networks (CNNs) using a delayed impulsive control approach. To address the disturbances caused by parameter uncertainty and the flexibility of impulsive delays, the concept of average impulsive delay (AID) and average impulsive interval (AII) were utilized to handle the delays as a whole. Under the condition that the norms of uncertain parameters are bounded, the synchronization criteria for uncertain CNNs were derived based on linear matrix inequalities (LMIs). Specifically, we relaxed the constraints on the delay in the impulsive control inputs, thus allowing it to flexibly vary without being bound by some conditions, which provides a broader applicability compared to most existing results. Additionally, the results show that delayed impulses can facilitate the synchronization of uncertain CNNs. Finally, the validity of the theoretical results was verified through a numerical example.

    Citation: Biwen Li, Yujie Liu. Average-delay impulsive control for synchronization of uncertain chaotic neural networks with variable delay impulses[J]. Mathematical Biosciences and Engineering, 2025, 22(6): 1382-1398. doi: 10.3934/mbe.2025052

    Related Papers:

    [1] Biwen Li, Qiaoping Huang . Synchronization of time-delay systems with impulsive delay via an average impulsive estimation approach. Mathematical Biosciences and Engineering, 2024, 21(3): 4501-4520. doi: 10.3934/mbe.2024199
    [2] Luyao Li, Licheng Fang, Huan Liang, Tengda Wei . Observer-based event-triggered impulsive control of delayed reaction-diffusion neural networks. Mathematical Biosciences and Engineering, 2025, 22(7): 1634-1652. doi: 10.3934/mbe.2025060
    [3] Haitao Zhu, Xinrui Ji, Jianquan Lu . Impulsive strategies in nonlinear dynamical systems: A brief overview. Mathematical Biosciences and Engineering, 2023, 20(2): 4274-4321. doi: 10.3934/mbe.2023200
    [4] A. Vinodkumar, T. Senthilkumar, S. Hariharan, J. Alzabut . Exponential stabilization of fixed and random time impulsive delay differential system with applications. Mathematical Biosciences and Engineering, 2021, 18(3): 2384-2400. doi: 10.3934/mbe.2021121
    [5] Linni Li, Jin-E Zhang . Input-to-state stability of stochastic nonlinear system with delayed impulses. Mathematical Biosciences and Engineering, 2024, 21(2): 2233-2253. doi: 10.3934/mbe.2024098
    [6] Yilin Tu, Jin-E Zhang . Event-triggered impulsive control for input-to-state stability of nonlinear time-delay system with delayed impulse. Mathematical Biosciences and Engineering, 2025, 22(4): 876-896. doi: 10.3934/mbe.2025031
    [7] Xueyan Yang, Xiaodi Li, Qiang Xi, Peiyong Duan . Review of stability and stabilization for impulsive delayed systems. Mathematical Biosciences and Engineering, 2018, 15(6): 1495-1515. doi: 10.3934/mbe.2018069
    [8] Zixiao Xiong, Xining Li, Ming Ye, Qimin Zhang . Finite-time stability and optimal control of an impulsive stochastic reaction-diffusion vegetation-water system driven by Lˊevy process with time-varying delay. Mathematical Biosciences and Engineering, 2021, 18(6): 8462-8498. doi: 10.3934/mbe.2021419
    [9] Yawei Shi, Hongjuan Wu, Chuandong Li . Constrained hybrid control for parametric uncertainty systems via step-function method. Mathematical Biosciences and Engineering, 2022, 19(11): 10741-10761. doi: 10.3934/mbe.2022503
    [10] Ganesh Kumar Thakur, Sudesh Kumar Garg, Tej Singh, M. Syed Ali, Tarun Kumar Arora . Non-fragile synchronization of BAM neural networks with randomly occurring controller gain fluctuation. Mathematical Biosciences and Engineering, 2023, 20(4): 7302-7315. doi: 10.3934/mbe.2023317
  • This paper investigated the synchronization issue of uncertain chaotic neural networks (CNNs) using a delayed impulsive control approach. To address the disturbances caused by parameter uncertainty and the flexibility of impulsive delays, the concept of average impulsive delay (AID) and average impulsive interval (AII) were utilized to handle the delays as a whole. Under the condition that the norms of uncertain parameters are bounded, the synchronization criteria for uncertain CNNs were derived based on linear matrix inequalities (LMIs). Specifically, we relaxed the constraints on the delay in the impulsive control inputs, thus allowing it to flexibly vary without being bound by some conditions, which provides a broader applicability compared to most existing results. Additionally, the results show that delayed impulses can facilitate the synchronization of uncertain CNNs. Finally, the validity of the theoretical results was verified through a numerical example.



    In recent decades, neural networks (NNs) have received much attention for their ability to simulate complex dynamics and are widely applied in fields such as nonlinear programming and image processing[1,2,3,4,5,6,7]. Some NNs exhibit a chaotic behavior, thus adding to the complexity of the system. Chaotic attractors have been discovered in three-dimensional autonomous networks [1,2], and research indicates that chaotic behavior in the mammalian brain is related to associative memory and neuronal synchronization [3,4]. Neuronal synchronization is crucial for information processing. Additionally, the chaotic characteristics also provide new avenues for information encryption [5], with their unpredictability and sensitivity making them effective tools for secure communication. As technology advances, the research on chaotic neural networks (CNNs) will continue to deepen and expand.

    The study of chaotic synchronization has received a lot of attention due to its important applications in encryption, financial prediction, and secure communication, and other fields [5,8,9,10,11]. Over the past two decades, impulsive control (IC) has been widely used as an effective method for system stabilization and chaotic synchronization [12,13,14,15]. It has been shown that CNNs can achieve synchronization under IC. The advantage is that discrete control inputs are only applied at specific moments, thus simplifying the controller structure and reducing communication stress and control costs. In addition, impulsive control has the several significant advantages. First, it enables fast system synchronization with less control energy, thus improving the control efficiency. Second, this control strategy is highly adaptable to the dynamic characteristics of the system and can better cope with uncertainties (such as random noise and unknown parameters) and nonlinear factors in complex systems. This efficient and economical strategy has been extensively applied in complex systems such as signal processing, financial market analyses, and neural networks [16,17,18,19]. With the technological advancements, the application prospects of impulsive control in chaotic synchronization are expected to become more broader.

    In practical systems, the input of impulsive signals and the sampling of system states are usually affected by time delays. In recent years, the research on time-delay impulsive differential systems has gradually gained attention, with a continuous emergence of relevant results [16,20,21,22,23,24]. For instance, researchers have explored the existence and uniqueness of solutions for time-delay impulsive systems [25,26], stability criteria based on the Lyapunov method [20,21], and chaos synchronization under time-delay impulsive control [14,21,22]. However, there are still two aspects that need further investigation in the context of time-delay impulsive synchronization control. First, the effect of impulse the control input on the synchronization performance has not been fully clarified [21]. Second, most existing studies assume that the impulse delay is either a fixed constant or subject to strict constraints, which may lead to conservatism in the research outcomes[20,24]. In previous studies, the authors typically treated the time delay as a fixed constant or assumed that the delay varied within a specific range but did not exceed a predefined upper bound [27]. While this approach offers some conveniences in the theoretical analysis, the dynamic nature of impulsive delays (i.e., delays that vary with the timing of the impulsive) often makes such assumptions overly simplistic in practical applications. This might restrict the scope of application of the research outcomes in the real world. In reality, impulsive delays are not static but dynamically change with different impulsive triggering moments. This dynamic characteristic makes traditional fixed-delay approaches inadequate for the effective handling of impulsive delays. Therefore, it would be helpful to more effectively address the impulsive delay problem if it were treated as a whole. Recently, a new concept called the average impulsive delay (AID) has emerged. By using the idea of an AID, we can effectively mitigate the impact of varying delays at different impulsive moments and handle impulsive delays from a holistic perspective.

    On the other hand, the system parameters are often affected by factors such as modeling inaccuracies, external disturbances, and measurement errors during the actual operation, thereby exhibiting uncertainty. When the model is subjected to disturbances in real-world scenarios (including nonlinear coefficients, external environmental interferences, periodic fluctuations, stochastic noise, and time-varying uncertainties), synchronization errors may fail to converge and even diverge, thus leading to system instability. Higher levels of uncertainty may trigger bifurcations (such as Hopf bifurcations), which can disrupt the synchronized state. Moreover, measurement errors and noise can further amplify these errors, thus causing the system to deviate from its intended trajectory. Therefore, designing appropriate control strategies to suppress uncertainties is crucial to ensure that the system remains stably synchronized in the presence of parameter perturbations. In recent years, the issue of parameter uncertainty has gradually become a research hotspot, which has attracted the attention of many scholars [28,29,30]. For example, in [28], the author proposed a method based on an improved Lyapunov-Krasovskii functional combined with linear matrix inequalities (LMIs) techniques, which established a new stability criteria for systems with parameter uncertainties. In previous studies [31], the author investigated the robust stability of nonlinear systems with mixed delays and parameter uncertainties. Moreover, due to the high sensitivity of chaotic dynamic systems to parameter changes, even slight uncertainties can have a significant effect on the overall behavior of the system. Therefore, achieving synchronization of chaotic systems under parameter uncertainties remains a challenging problem.

    Based on the brief summary and discussion above, the purpose of this paper is to study the synchronization problem of CNNs with parameter uncertainties using the delayed impulse control method. The following are the main contributions of this paper:

    (1) A delayed impulse control method based on an AID and an average impulsive interval (AII) is proposed to address the synchronization problem of uncertain CNNs. This method effectively relaxes the constraints on the delay of the impulse control inputs, thus making it more flexible than those in most existing studies.

    (2) Several sufficient conditions for reaching the synchronization of CNNs with parametric uncertainties are derived using LMIs. In comparison to existing research, the obtained synchronization conditions are more general and significantly reduce conservatism.

    (3) The effectiveness of the proposed method is verified through a theoretical analysis and numerical simulations. The results show that delayed impulses can significantly promote the synchronization of uncertain CNNs, which provide additional insights for research in this area.

    Notations: Let Rn represent the n-dimensional Euclidean space, and Rn×m represent the n×m-dimensional Euclidean space. The notation Z+ denotes the set of positive integers, and R stands for the set of real numbers. The notation Y>0 indicates that the matrix Y is positive definite. The symbol denotes the Euclidean norm for matrices or vectors. For any matrix Λ, ΛT denotes its transpose and Λ1 denotes its inverse. The symbol denotes a symmetric matrix block. Matrix I is the identity matrix.

    Let the following uncertain CNNs serve as the drive system:

    ˙Q(t)=(F+F(t))Q(t)+(E+E(t))φ(Q(t))+J, (2.1)

    where Q(t)=(Q1(t),...,Qn(t))T denotes the state variable of the drive system, ˙Q(t) stands for the derivative of the state variable, E denotes the neuron connection weight matrix, F=diag(f1,...,fn) denotes a diagonal matrix, where fi>0,iZ+, E(t) and F(t) denote time-varying parameter uncertain matrices, J stands for the external input, and φ is a nonlinear activation function.

    Consider the following system as the response system for uncertain CNNs (2.1) with parameter uncertainties:

    {˙R(t)=(F+F(t))R(t)+(E+E(t))φ(R(t))+J,ttk,R(t)=U(t),t=tk, (2.2)

    where R(t)=(R1(t),...Rn(t)) stands for the state variable of system (2.2), R(t)=R(tk)R(tk), and the impulsive sequence {tk}kZ+ is strictly increasing, which satisfies t1<t2<<tk<,limktk=+. We assume that system (2.2) is right continuous at every instant t=tk, (i.e., R(t+k)=R(t)). In this paper, we design U(t) as follows:

    U(tk)=Ψς((tkδk))ς(tk),kZ+, (2.3)

    where ΨRn×n is the impulsive gain matrix, and δk is the time delay in impulsive. ς(t)=R(t)Q(t) is defined as the synchronization error of systems (2.1) and (2.2). Then,

    ς(tk)=R(tk)Q(tk)=R(tk)+R(tk)Q(tk)=R(tk)+Ψς((tkδk))ς(tk)Q(tk)=R(tk)Q(tk)+Ψς((tkδk))ς(tk)=ς(tk)+Ψς((tkδk))ς(tk)=Ψς((tkδk)). (2.4)

    Then, the error system is obtained by subtracting system (2.1) from system (2.2) as follows:

    {˙ς(t)=(F+F(t))ς(t)+(E+E(t))φ(ς(t)),ttk,ς(tk)=Ψς((tkδk)), (2.5)

    where φ(ς(t))=φ(R(t))φ(Q(t)), and we assume that the delayed impulsive δk satisfies tk<δk<tk+1.

    The following are some necessary assumptions, definitions, and lemmas presented in this paper.

    Assumption 1 ([11]).Suppose that the activation function φ satisfies the Lipschitz condition, that is, there exists a positive constant ι, such that

    |φj(z)φj(r)|ι|zr|, (2.6)

    where z,rR, j=1,...,n.

    Definition 1 ([6]). The method for calculating the AII Iθ between two consecutive impulsive signals is defined as follows:

    Iθ=lim supt+tt0Mξ(t,t0), (2.7)

    where Mξ(t,t0) represents the count of impulses within the impulse sequence {tk} that occur in the interval (t0,t).

    Definition 2 ([7]). The AID ˉδ for the impulse delay sequence {δk}kZ+ is defined as follows:

    ˉδ=lim inft+δ1+δ2+...+δMξ(t,t0)Mξ(t,t0), (2.8)

    where Mξ(t,t0) represents the count of impulses within the impulse sequence {tk} that occur in the interval (t0,t).

    Definition 3 ([22]) Generally, if uncertain CNNs (2.1) and (2.2) satisfy the condition limt+R(t)Q(t)=0, then they are considered synchronized, where R(t) and Q(t) represent the solutions to systems (2.1) and (2.2), respectively. Consequently, by the definition of synchronization, if the system error ς(t)=R(t)Q(t) satisfies the condition

    limt+ς(t)=0,

    then, we can conclude that systems (2.1) and (2.2) have achieved synchronization.

    Lemma 1 ([11]).Let ε>0; for any matrix G1,G2Rn×n, there exist a positive define matrix D>0, such that

    GT1G2+GT2G1εGT1DG1+1εGT2D1G2.

    Lemma 2 ([32]).Let u1 and u2 be real vectors with appropriate dimensions, and S and A(t) be real matrices with the appropriate dimensions, A(t)γ; for α>0, the following inequality holds:

    2uT1STA(t)u2αuT1STSu1+γ2αuT2u2.

    Lemma 3 ([33]).Give the following matrix:

    E=(E11E12E21E22),

    where ET11=E11,ET12=E21,ET22=E22; then, E<0 is equivalent to any of the following conditions:

    (1) E11<0, and E22ET12E111E12<0.

    (2) E22<0, and E11E12E122ET12<0.

    Remark 1. Definition 1 and [34] define the AII in different ways. Definition 1 in [6] uses the upper limit Iθ=lim supt+tt0Mξ(t0,t) to describe the long-term average behavior of the impulsive sequence. It is widely applicable, especially for analyzing non-periodic or highly stochastic impulsive sequences. In contrast, reference [34] constrains the number of impulses in any time interval using the upper and lower bounds t2t1IθM0Mξ(t2,t1)t2t1Iθ+M0. Although this provides a more precise local control, it is more conservative and has a narrower scope of application. Definition 1 is less conservative because it imposes weaker constraints on the impulsive sequence, thus making it applicable for a broader range of scenarios while maintaining the universality of the theoretical analysis. It is particularly well-suited to study the long-term behavior and global stability of the systems.

    Remark 2. Definitions 1 and 2 describe the characteristics of the impulsive sequence from the different perspectives of impulse intervals and impulse delays, respectively. Note that since both AII and AID are related to Mξ(t,t0), we can derive the following relationship:

    lim inft+δ1+δ2++δMξ(t,t0)tt0=ˉδIθ.

    Remark 3. We define two sets of time sequences and , which correspond to {tk}kZ+ and {δk}kZ+ that satisfy (2.7) and (2.8), respectively. To simplify the expression, we denote the collection of sequences that satisfy {tk}kZ+ and {δk}kZ+ as H({tk},{δk})kZ+. We denote (Ψ,H) as an impulsive controller aimed at synchronizing the uncertain CNNs (2.1) and (2.2). Additionally, we term this control strategy as the average-delay impulsive control.

    In this section, the average-delay impulsive control method will be used to study the chaotic synchronization between systems (2.1) and (2.2). By the definitions of AII and AID, we have relaxed the restrictions on impulse delay and have derived the synchronization criteria between systems (2.1) and (2.2) based on LMIs. The results show that impulsive delay promotes the synchronization of the two uncertain CNNs.

    Theorem 1. Assume that Assumption 1 holds and under the average-delayed impulse controller (Ψ,H). If there exists positive constants π>0, λ>0, and positive definite matrices N>0,D>0 such that

    (FTNNF+(α1+α2)N2+γ21α1In+γ22α2L2+LDL+NED1ETNπNETND)<0, (3.1)
    (λNΨTNN)<0, (3.2)

    and

    lnλ+π(Iθˉδ)<0, (3.3)

    where L=diag(ι1,...,ιn), then, uncertain CNNs (2.1) and (2.2) can reach synchronization.

    Proof. We choose a nonnegative Lyapunov function for the error system (2.5):

    V(t)=ςT(t)Nς(t), (3.4)

    where N>0; then, based on Lemmas 1 and 2, for t[tk1,tk),kZ+, the right hand derivative of the Lyapunov function V(t) along the error system (2.5) is calculated as follows:

    ˙V(t)=˙ςT(t)Nς(t)+ςT(t)N˙ς(t)=[(F+F(t))ς(t)+(E+E(t))φ(ς(t))]TNς(t)+ςT(t)N[(F+F(t))ς(t)+(E+E(t))φ(ς(t))]=ςT(t)[FTNNF]ς(t)+ςT(t)[(F(t))TNNF(t)]ς(t)+φT(ς(t))ETNς(t)+ςT(t)NEφ(ς(t))+φT(ς(t))(E(t))TNς(t)+ςT(t)NE(t)φ(ς(t)). (3.5)

    According to (2.6), Lemmas 1 and 2, we can derive the following:

    ςT(t)[(F(t))TNN(F(t))]ς(t)=2ςT(t)NF(t)ς(t)α1ςT(t)N2ς(t)+γ21α1ςT(t)ς(t),
    φT(ς(t))ETNς(t)+ςT(t)NEφ(ς(t))φT(ς(t))Dφ(ς(t))+ςT(t)NED1ETNς(t)ςT(t)LDLς(t)+ςT(t)NED1ETNς(t),
    φT(ς(t))(E(t))TNς(t)+ςT(t)NE(t)φ(ς(t))α2ςT(t)N2ς(t)+1α2φT(ς(t))(E(t))T(E(t))φ(ς(t))α2ςT(t)N2ς(t)+γ22α2ςT(t)L2ς(t).

    From Lemma 3 and (3.1), one can obtain the following:

    FTNNF+(α1+α2)N2+γ21α1In+γ22α2L2+LDL+NED1ETNπN<0. (3.6)

    Then, by substituting the result of the simplification above into Eq (3.5), one can obtain the following:

    ˙V(t)ςT(t)(FTNNF+(α1+α2)N2+γ21α1In+γ22α2L2+LDL+NED1ETNπN)ς(t)ςT(t)πNς(t)πV(t). (3.7)

    From (3.7), we derive the following:

    V(t)V(tk1)eπ(ttk1). (3.8)

    Additionally, from (3.2), it can be derived that at the impulsive instant t=tk, the following equation holds:

    V(tk)=ςT(tk)Nς(tk)ςT((tkδk))ΨTNΨς((tkδk))=ςT((tkδk))ΨTNN1NΨς((tkδk))<ςT((tkδk))λNς((tkδk))=λV((tkδk)). (3.9)

    The following analysis is based on (3.8) and (3.9).

    Next, we will prove that the following inequality holds, for pZ+,t[tp1,tp):

    V(t)λp1exp[πp1j=0δj]eπ(tt0)V0, (3.10)

    where V0=ς(t0)TNς(t0), and ς(t)=ς(t,t0,ς(t0)) represents the solution of the error system (2.5) through the initial value ς(t0). Note that since δ0 stands for the impulsive delay at the impulsive moment t=0, there is no control input at t=0. Therefore, when j=0, δ0=0.

    For t[t0,t1), the following is derived from (3.8),

    V(t)eπ(tt0)V0. (3.11)

    Therefore, it can be verified that (3.10) holds when p=1. Next, we assume that (3.10) also holds for p=q, i.e., qZ+,t[tq1,tq),

    V(t)λq1exp[πq1j=0δj]eπ(tt0)V0, (3.12)

    when p=q+1, from tk1<δk<tk,kZ+; then, tqδq[tq1,tq), and according to (3.9) and (3.10),

    V(tq)=λV((tqδq))λλq1exp[πq1j=0δj]eπ(tqδqt0)V0=λqexp[πqj=0δj]eπ(tqt0)V0. (3.13)

    For t[tq,tq+1), according to (3.8), we can get that V(t)V(tq)eπ(ttq); substituting (3.13) into it, we have the following:

    V(t)λqexp[πqj=0δj]eπ(tqt0)V0eα(ttq)=λqexp[πqj=0δj]eπ(tt0)V0, (3.14)

    i.e., when p=q+1, (3.10) also holds. Therefore, through mathematical induction, we can deduce that (3.10) holds, for pZ+. Furthermore, since the condition {tk}kZ+ and {δk}kZ+, it follows that

    V(t)λMξ(t,t0)exp[πMξ(t,t0)j=0δj]eπ(tt0)V0=exp[Mξ(t,t0)lnλπMξ(t,t0)j=0δj]eπ(tt0)V0=exp[(Mξ(t,t0)lnλπMξ(t,t0)j=0δjMξ(t,t0)/tt0Mξ(t,t0))×(tt0)]eπ(tt0)V0=exp[(Mξ(t,t0)lnλπMξ(t,t0)j=0δjMξ(t,t0)/tt0Mξ(t,t0)+π)×(tt0)]V0. (3.15)

    Then, based on the definitions of the AII, AID, and (3.3), we can obtain the following:

    lim supt+[(Mξ(t,t0)lnλπMξ(t,t0)j=0δjMξ(t,t0)/tt0Mξ(t,t0)+π)×(tt0)]V0lnλπˉδ+πIθIθ<0. (3.16)

    Then, it follows from (3.15) and (3.16) that

    limt+V(t)=0.

    Taking (3.4) into account, it follows that limt+ς(t)TNς(t)=0. As a result, the uncertain CNNs systems (2.1) and (2.2) achieves synchronization. The proof is complete.

    Remark 4. In previous studies (such as references [14,21,22]), the time delay of the impulsive signal (i.e., signal delay) is usually supposed that the impulse delay is a fixed constant or subject to strict constraints. This assumption makes the existing research criteria relatively conservative in dealing with impulsive time delay problems. However, in practical systems, the time delay of the impulsive signal is not always fixed or strictly limited; on the contrary, it can vary flexibly. This flexibility is common in practical applications, though previous studies have not fully considered this aspect. Based on this, this paper employs the concept of an AID. This means that the time delay in the impulse control input can flexibly exist between two consecutive impulse moments. Compared with the AID concept in reference [7], this paper considers a more generalized lower-bound form of the AID concept. Additionally, this paper has employed the LMIs method to verify and analyze the effectiveness of the proposed controller.

    Remark 5. To achieve synchronization between uncertain CNNs (2.1) and (2.2), impulsive control inputs need to be continuously implemented. This means that to ensure these two systems can synchronize, Iθ should not be set too large. According to condition (3.3), we know that Iθ<ˉδlnλπ. Because of the time delay in the impulse control inputs, we find that Iθ can be larger than the value allowed without delay (i.e., Iθ<lnλπ). Intuitively, it is observed that within each impulse interval (tk1,tk), the error ς(t) may increase due to the absence of impulsive control inputs. This implies that under the condition tk1<tkδktk, the inequality ς((tkδk))ς(tk) might be satisfied. Therefore, taking this into account, we can infer that the time delay of the impulse control input may positively affect the synchronization of systems (2.1) and (2.2) (this will be verified in the numerical examples section).

    In particular, if both the impulse interval and the delay of the impulse control input are set to constant values, (i.e., tktk1=I and δk=δ, kZ+), then the following corollary can be drawn.

    Corollary 1. Suppose Assumption 1 is valid. Set the impulse interval I>0, the impulse delay δ0, and the constants π>0,λ>0; assuming there exists positive definite matrices N>0 and D>0, which satisfy

    ΨTNΨeπ(Iδ)N<0,

    and (3.1) and (3.2) hold, then

    ΨTNN1NΨλN<0,

    and

    FTNNF+(α1+α2)N2+γ21α1In+γ22α2L2+LDL+NED1ETNπN<0,

    will hold. Then uncertain CNNs (2.1) and (2.2) can reach synchronization under the impulse controller (Ψ,{I,δ}).

    In this section, the validity of the theoretical results under the proposed average-delay impulsive control strategy is verified by providing an example. We select the three-neuron uncertain CNNs (4.1) as the drive system and (4.2) as the corresponding response system. Their forms are as follows:

    ˙Q(t)=(F+F(t))Q(t)+(E+E(t))φ(Q(t))+J. (4.1)

    The corresponding response system is as follows:

    {˙R(t)=(F+F(t))R(t)+(E+E(t))φ(R(t))+J,ttk,R(t)=U(t),t=tk, (4.2)

    where the uncertain CNNs (4.1) has the initial condition Q0=(0.75,1.5,0.3). The behaviors of the uncertain CNNs (4.1) with the initial value Q0 is shown in Figure 1, and the evolution of error norm ς(t) without control is shown in Figure 2.

    Figure 1.  The behaviors of the uncertain CNNs (4.1).
    Figure 2.  Evolution of the error norm ς(t) without control.

    The response system (4.2) has the initial condition R0=(2.5,2.1,2.6), and the nonlinear activation function

    φ(Q(t))=[φ1(Q1(t)),φ2(Q2(t)),φ3(Q3(t))]T,Q(t)R3,

    takes the form

    φi(Qi(t))=12(|Qi(t)+1||Qi(t)1|),i=1,2,3.

    Moreover,

    F=[100010001],E=[1.21.601.210.902.21.5],
    F(t)=0.13cos(t)[100010001],E(t)=0.1cos(t)[111111111],J=0.

    It is evident that, the norm of the time-varying parameter uncertainty matrices F(t) and E(t) are bounded, under the above conditions, (i.e., F(t)0.13,E(t)0.2). Additionally, we can calculate to obtain L=diag(ι1,ι2,ι3)=I. Select the parameters π=3.9,λ=0.4, and the impulsive gain matrix Ψ=0.6I. Moreover, with the help of MATLAB's Toolbox LMI, it can be verified that LMIs (3.1) and (3.2) in Theorem 1 are satisfied. Additionally, the following are the relevant feasible solutions:

    N=[0.30750.00700.04720.00700.33260.02060.04720.02060.2626],
    D=[0.81780.00650.19970.00650.94360.03330.19970.03330.6372].

    In this simulation example, we set the impulse sequence tk=5k, kZ+. When there is a delay in the impulsive control input, the delay ˉδ needs to satisfy the condition ˉδ>Iθ+(lnλ)/π4.77. Under this condition, the drive-response systems (4.1) and (4.2) can achieve a synchronized state, which can be observed from the evolutionary trajectory of the error system in Figure 3, and the red solid line in Figure 4.

    Figure 3.  Evolution of the error ςi(t) under average-delay impulsive control.
    Figure 4.  Evolution of the error norm ς(t) under different δk.

    Specifically, the designed delayed impulsive sequence is δk=4.85+((1)k)/20+0.04|cos(k2)|, and it can be verified that ˉδ4.85, which satisfies the condition (3.3) in Theorem 1. However, when there is no delay in the impulsive sequence, that is, δk=0, the uncertain CNNs (4.1) and (4.2) cannot achieve synchronization, which can be observed from the blue dashed line in Figure 4. Under this condition, we can calculate that π(Iθˉδ)+lnλ0 does not satisfy the condition (3.3) in Theorem 1, which further confirms the positive effect of a delayed impulsive on achieving synchronization.

    Moreover, if other parameters remain constant, then increasing the impulse control intensity results in a faster convergence rate of the system error. Similarly, if the impulse delay is increased while still satisfying (3.3) in Theorem 1, then the rate of stabilization of the error system will also be faster, as shown in Figure 5 (ˆΨ=diag{0.4,0.4,0.4},^δk=4.86+((1)k)/20+0.04|cos(k2)|). On the contrary, reducing the impulse gain and decreasing the impulse delay may lead to an instability of the error system (i.e., divergence), as shown in Figure 6 (ˇΨ=diag{0.67,0.6,0.6},ˇδk=3+((1)k)/20+0.04|cos(k2)|). This demonstrates that the control strategy designed in this paper is effective.

    Figure 5.  Evolution of the error under the control parameter ˆΨ and ˆδk.
    Figure 6.  Evolution of the error under the control parameter ˇΨ and ˇδk.

    In this paper, the synchronization problem of uncertain CNNs was investigated via the delayed impulsive control technology. To address the variable delays in the impulsive control inputs, an average-delay impulsive control strategy was utilized. By utilizing the concepts of AII and AID, some sufficient conditions were derived based on LMIs for the synchronization of uncertain CNNs. The results of the study show that the presence of impulsive delays can effectively prompt the stabilization of the error system. Future research will consider the synchronization problem of delayed uncertain CNNs under the delayed impulse control.

    Biwen Li: Sriting–review, supervision, validation and formal analysis; Yujie Liu: Writing–review, writing–original draft, methodology, editing and software. Both authors have read and approved the final version of the manuscript for publication.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    The authors declare there is no conflict of interest.



    [1] A. Das, P. Das, A. B. Roy, Chaos in a three-dimensional general model of neural network, Int. J. Bifurcation Chaos, 12 (2002), 2271–2281. https://doi.org/10.1142/S0218127402005820 doi: 10.1142/S0218127402005820
    [2] Y. Horikawa, H. Kitajima, H. Matsushita, Fold-pitchfork bifurcation, arnold tongues and multiple chaotic attractors in a minimal network of three sigmoidal neurons, Int. J. Bifurcation Chaos, 28 (2018), 1850123. https://doi.org/10.1142/S0218127418501237 doi: 10.1142/S0218127418501237
    [3] A. Babloyantz, C. Lourenço, Brain chaos and computation, Int. J. Neural Syst., 7 (1996), 461–471. https://doi.org/10.1142/S0129065796000440
    [4] P. Fries, J. H. Reynolds, A. E. Rorie, R. Desimone, Modulation of oscillatory neuronal synchronization by selective visual attention, Science, 291 (2001), 5508. https://doi.org/10.1126/science.1055465 doi: 10.1126/science.1055465
    [5] A. Khadra, X. Z. Liu, X. Shen, Impulsively synchronizing chaotic systems with delay and applications to secure communication, Automatica, 41 (2005), 1491–1502. https://doi.org/10.1016/j.automatica.2005.04.012 doi: 10.1016/j.automatica.2005.04.012
    [6] N. Wang, X. Li, J. Lu, F. E. Alsaadi, Unified synchronization criteria in an array of coupled neural networks with hybrid impulses, IEEE Trans. Neural Networks Learn. Syst., 101 (2018), 25–32. https://doi.org/10.1016/j.neunet.2018.01.017 doi: 10.1016/j.neunet.2018.01.017
    [7] B. Jiang, J. Lu, J. Lou, J. Qiu, Synchronization in an array of coupled neural networks with delayed impulses: average impulsive delay method, Neural Networks, 121 (2020), 452–460. https://doi.org/10.1109/TNNLS.2019.2927249 doi: 10.1109/TNNLS.2019.2927249
    [8] M. Shafiq, I. Ahmad, Robust synchronization of four-dimensional chaotic finance systems with unknown parametric uncertainties, Automatika, 65 (2024), 217–234. https://doi.org/10.1080/00051144.2023.2295204 doi: 10.1080/00051144.2023.2295204
    [9] W. He, T. Luo, Y. Tang, W. Du, Y. C. Tian, F. Qian, Secure communication based on quantized synchronization of chaotic neural networks under an event-triggered strategy, IEEE Trans. Neural Networks Learn. Syst., 31 (2020), 3334–3345. https://doi.org/10.1109/TNNLS.2019.2943548 doi: 10.1109/TNNLS.2019.2943548
    [10] J. J. Xiong, G. B. Zhang, J. X. Wang, T. H. Yan, Improved sliding mode control for finite-time synchronization of nonidentical delayed recurrent neural networks, IEEE Trans. Neural Networks Learn. Syst., 31 (2020), 2209–2216. https://doi.org/10.1109/TNNLS.2019.2927249 doi: 10.1109/TNNLS.2019.2927249
    [11] J. G. Lu, G. Chen, Global asymptotical synchronization of chaotic neural networks by output feedback impulsive control: an LMI approach, Chaos, Solitons Fractals, 41 (2009), 2293–2300. https://doi.org/10.1016/j.chaos.2008.09.024 doi: 10.1016/j.chaos.2008.09.024
    [12] B. Li, Y. Liu, Quasi-synchronization of nonlinear systems with parameter mismatch and time-varying delays via event-triggered impulsive control, AIMS Math., 10 (2025), 3759–3778. https://doi.org/10.3934/math.2025174 doi: 10.3934/math.2025174
    [13] Y. Tang, X. Wu, P. Shi, F. Qian, Input-to-state stability for nonlinear systems with stochastic impulses, Automatica, 13 (2020), 108766. https://doi.org/10.1016/j.automatica.2019.108766 doi: 10.1016/j.automatica.2019.108766
    [14] Z. Xu, D. Peng, X. Li, Synchronization of chaotic neural networks with time delay via distributed delayed impulsive control, Neural Networks, 118 (2019), 332–337. https://doi.org/10.1016/j.neunet.2019.07.002 doi: 10.1016/j.neunet.2019.07.002
    [15] Q. Tang, S. Qu, C. Zhang, Z. Tu, Y. Cao, Effects of impulse on prescribed-time synchronization of switching complex networks, Neural Networks, 174 (2024), 106248. https://doi.org/10.1016/j.neunet.2024.106248 doi: 10.1016/j.neunet.2024.106248
    [16] W. Zhu, D. Wang, L. Liu, G. Feng, Event-based impulsive control of continuous time dynamic systems and its application to synchronization of memristive neural networks, IEEE Trans. Neural Networks Learn., 29 (2018), 3599–3609. https://doi.org/10.1109/TNNLS.2017.2731865 doi: 10.1109/TNNLS.2017.2731865
    [17] Z. W. Liu, G. Wen, X. Yu, Z. H. Guan, T. Huang, Delayed impulsive control for consensus of multiagent systems with switching communication graphs, IEEE Trans. Cybern., 50 (2020), 3045–3055. https://doi.org/10.1109/TCYB.2019.2926115 doi: 10.1109/TCYB.2019.2926115
    [18] X. Lv, X. Li, J. Cao, M. Perc, Dynamical and static multisynchronization of coupled multistable neural networks via impulsive control, Neural Networks Learn. Syst., 529 (2018), 6062–6072. https://doi.org/10.1109/TNNLS.2018.2816924 doi: 10.1109/TNNLS.2018.2816924
    [19] J. Shao, H. Jiang, S. K. Nguang, H. Shen, Impulsive synchronization of coupled delayed neural networks with actuator saturation and its application to image encryption, Neural Networks, 128 (2020), 158–171. https://doi.org/10.1016/j.neunet.2020.05.016 doi: 10.1016/j.neunet.2020.05.016
    [20] W. H. Chen, W. X. Zheng, Exponential stability of nonlinear time delay systems with delayed impulse effects, Automatica, 47 (2011), 1075–1083. https://doi.org/10.1016/j.automatica.2011.02.031 doi: 10.1016/j.automatica.2011.02.031
    [21] X. Li, S. Song, J. Wu, Exponential stability of nonlinear systems with delayed impulses and applications, IEEE Trans. Autom. Control, 64 (2019), 4024–4034. https://doi.org/10.1109/TAC.2019.2905271 doi: 10.1109/TAC.2019.2905271
    [22] B. Jiang, J. Lou, J. Lu, K. Shi, Synchronization of chaotic neural networks: average-delay impulsive control, IEEE Trans. Neural Networks Learn. Syst., 33 (2022), 6007–6012. https://doi.org/10.1109/TNNLS.2021.3069830 doi: 10.1109/TNNLS.2021.3069830
    [23] F. Fan, Y. Xiao, K. Shi, H. Wen, Y. Zhao, μ-synchronization of coupled neural networks with hybrid delayed and non-delayed impulsive effects, Chaos, Solitons Fractals, 173 (2023), 113620. https://doi.org/10.1016/j.chaos.2023.113620 doi: 10.1016/j.chaos.2023.113620
    [24] X. Li, S. Song, Stabilization of delay systems: delay-dependent impulsive control, IEEE Trans. Autom. Control, 62 (2017), 406–411. https://doi.org/10.1109/TAC.2016.2530041 doi: 10.1109/TAC.2016.2530041
    [25] G. Ballinger, X. Liu, Existence, uniqueness and boundedness results for impulsive delay differential equations, Dyn. Continuous Discrete Impuls. Syst., 74 (2000), 71–93. https://doi.org/10.1080/00036810008840804 doi: 10.1080/00036810008840804
    [26] X. Liu, G. Ballinger, Existence and continuability of solutions for differential equations with delays and state-dependent impulses, Nonlinear Anal. Theory Methods Appl., 51 (2002), 633–647. https://doi.org/10.1109/TAC.2016.2530041 doi: 10.1109/TAC.2016.2530041
    [27] S. Dong, X. Liu, S. Zhong, K. Shi, H. Zhu, Practical synchronization of neural networks with delayed impulses and external disturbance via hybrid control, Neural Networks, 157 (2023), 54–64. https://doi.org/10.1016/j.neunet.2022.09.025 doi: 10.1016/j.neunet.2022.09.025
    [28] A. Wu, H. Liu, Z. Zeng, Observer design and H performance for discrete-time uncertain fuzzy-logic systems, IEEE Trans Cybern., 51 (2021), 2398–2408. https://doi.org/10.1109/TCYB.2019.2948562 doi: 10.1109/TCYB.2019.2948562
    [29] C. Ge, X. Liu, C. Hua, J. H. Park, Exponential synchronization of the switched uncertain neural networks with mixed delays based on sampled-data control, J. Franklin Inst., 54 (2022), 2259–2282. https://doi.org/10.1016/j.neunet.2014.02.008 doi: 10.1016/j.neunet.2014.02.008
    [30] C. Jin, Z. Wang, L. Gong, M. Xiao, G. Jiang, Quasi-synchronization of heterogeneous Lur'e networks with uncertain parameters and impulsive effect, Neurocomputing, 482 (2022), 252–263. https://doi.org/10.1016/j.neucom.2021.11.057 doi: 10.1016/j.neucom.2021.11.057
    [31] W. Huang, Q. Song, Z. Zhao, Y. Liu, F. E. Alsaadi, Robust stability for a class of fractional-order complex-valued projective neural networks with neutral-type delays and uncertain parameters, J. Franklin Inst., 450 (2021), 399–410. https://doi.org/10.1016/j.neucom.2021.04.046 doi: 10.1016/j.neucom.2021.04.046
    [32] F. Chen, W. Zhang, LMI criteria for robust chaos synchronization of a class of chaotic systems, Nonlinear Anal. Theory Methods Appl., 67 (2007), 3384–3393. https://doi.org/10.1016/j.na.2006.10.020 doi: 10.1016/j.na.2006.10.020
    [33] S. Boyd, L. El Ghaoui, E. Feron, V. Balakrishnan, Linear Matrix Inequalities in System and Control Theory, Philadelphia, PA, USA: SIAM, 1994. https://doi.org/10.1137/1.9781611970777
    [34] J. Lu, D. W. C. Ho, J. Cao, A unified synchronization criterion for impulsive dynamical networks, Automatica, 46 (2010), 1215–1221. https://doi.org/10.1016/j.automatica.2010.04.005 doi: 10.1016/j.automatica.2010.04.005
    [35] S. Arik, An improved robust stability result for uncertain neural networks with multiple time delays, Neural Networks, 54 (2014), 1–10. https://doi.org/10.1016/j.neunet.2014.02.008 doi: 10.1016/j.neunet.2014.02.008
    [36] H. Fan, J. Tang, K. Shi, Y. Zhao, H. Wen, Delayed impulsive control for μ-synchronization of nonlinear multi-weighted complex networks with uncertain parameter perturbation and unbounded delays, Mathematics, 599 (2023), 127484. https://doi.org/10.3390/math11010250 doi: 10.3390/math11010250
  • Reader Comments
  • © 2025 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(351) PDF downloads(34) Cited by(0)

Figures and Tables

Figures(6)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog