
In this article, the distributed optimization based on multi-agent systems was studied, where the global optimization objective of the optimization problem is a convex combination of local objective functions. In order to avoid continuous communication among neighboring agents, an event-triggering algorithm was proposed. Time delay was also considered in the designed algorithm. The triggering time of each agent was determined by the state measurement error, the state of its neighbors at the latest triggering instant and the exponential decay threshold. Some sufficient conditions for optimal consistency were obtained. In addition, Zeno-behavior in triggering time sequence was eliminated. Finally, a numerical simulation was given to prove the effectiveness of the proposed algorithm.
Citation: Run Tang, Wei Zhu, Huizhu Pu. Event-triggered distributed optimization of multi-agent systems with time delay[J]. Mathematical Biosciences and Engineering, 2023, 20(12): 20712-20726. doi: 10.3934/mbe.2023916
[1] | Wangming Lu, Zhiyong Yu, Zhanheng Chen, Haijun Jiang . Prescribed-time cluster practical consensus for nonlinear multi-agent systems based on event-triggered mechanism. Mathematical Biosciences and Engineering, 2024, 21(3): 4440-4462. doi: 10.3934/mbe.2024196 |
[2] | Mingxia Gu, Zhiyong Yu, Haijun Jiang, Da Huang . Distributed consensus of discrete time-varying linear multi-agent systems with event-triggered intermittent control. Mathematical Biosciences and Engineering, 2024, 21(1): 415-443. doi: 10.3934/mbe.2024019 |
[3] | Siyu Li, Shu Li, Lei Liu . Fuzzy adaptive event-triggered distributed control for a class of nonlinear multi-agent systems. Mathematical Biosciences and Engineering, 2024, 21(1): 474-493. doi: 10.3934/mbe.2024021 |
[4] | Duoduo Zhao, Fang Gao, Jinde Cao, Xiaoxin Li, Xiaoqin Ma . Mean-square consensus of a semi-Markov jump multi-agent system based on event-triggered stochastic sampling. Mathematical Biosciences and Engineering, 2023, 20(8): 14241-14259. doi: 10.3934/mbe.2023637 |
[5] | Qiushi Wang, Hongwei Ren, Zhiping Peng, Junlin Huang . Dynamic event-triggered consensus control for nonlinear multi-agent systems under DoS attacks. Mathematical Biosciences and Engineering, 2024, 21(2): 3304-3318. doi: 10.3934/mbe.2024146 |
[6] | Wenjing Wang, Jingjing Dong, Dong Xu, Zhilian Yan, Jianping Zhou . Synchronization control of time-delay neural networks via event-triggered non-fragile cost-guaranteed control. Mathematical Biosciences and Engineering, 2023, 20(1): 52-75. doi: 10.3934/mbe.2023004 |
[7] | Na Zhang, Jianwei Xia, Tianjiao Liu, Chengyuan Yan, Xiao Wang . Dynamic event-triggered adaptive finite-time consensus control for multi-agent systems with time-varying actuator faults. Mathematical Biosciences and Engineering, 2023, 20(5): 7761-7783. doi: 10.3934/mbe.2023335 |
[8] | Guodong Zhao, Haitao Li, Ting Hou . Survey of semi-tensor product method in robustness analysis on finite systems. Mathematical Biosciences and Engineering, 2023, 20(6): 11464-11481. doi: 10.3934/mbe.2023508 |
[9] | Dong Xu, Xinling Li, Weipeng Tai, Jianping Zhou . Event-triggered stabilization for networked control systems under random occurring deception attacks. Mathematical Biosciences and Engineering, 2023, 20(1): 859-878. doi: 10.3934/mbe.2023039 |
[10] | 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 |
In this article, the distributed optimization based on multi-agent systems was studied, where the global optimization objective of the optimization problem is a convex combination of local objective functions. In order to avoid continuous communication among neighboring agents, an event-triggering algorithm was proposed. Time delay was also considered in the designed algorithm. The triggering time of each agent was determined by the state measurement error, the state of its neighbors at the latest triggering instant and the exponential decay threshold. Some sufficient conditions for optimal consistency were obtained. In addition, Zeno-behavior in triggering time sequence was eliminated. Finally, a numerical simulation was given to prove the effectiveness of the proposed algorithm.
Due to its extensive applications, cooperative control of multi-agent systems (MASs) has received increasing attention for the past few years, such as consensus control [1,2], containment control [3], formation control [4] and resource allocation [5,6]. Distributed optimization problem (DOP), one of the hot topics of cooperative control, has been widely studied from different aspects. In DOP, each agent has its own local objective function, which is not available to other agents. The objective function is minimized by selecting an optimal action, where the objective function is defined as a convex combination of local objective functions. Optimization theories [7,8,9,10,11] provide the fundamental tools to address DOP.
In the field of DOP, there were some studies [12,13] about distributed consensus-based gradient methods for convex cost functions. Recently, DOP of continuous-time MASs have been widely researched [14,15,16]. In [14], the research was mainly based on nonuniform gradient gain and finite time convergence. In [15], the time-varying loss function was studied. A proportional-integral-differential algorithm was introduced in [16].
The majority of current DOP was studied using continuous-time control algorithms. This control strategy is relatively impractical and inefficient as it results in a wastage of energy. In practice, each agent often faces limited resources and, thus, is expected to update its control signals as infrequently as possible. In order to alleviate the communication load in MASs, a discrete-time control method known as event-triggered control has been developed. This control strategy is aperiodic and neighboring agents only need to communicate at specific time instants determined by pre-designed triggering conditions, the effectiveness of which was systematically illustrated in [17,18,19,20,21,22]. In [17], the research studied the event-triggered DOP for nonlinear MASs in undirected and connected communication networks. In [18], two event-triggered control protocols were proposed to solve convex DOP under the directed graph. Moreover, in [19], an event-based control protocol was proposed to solve non-convex DOP. An adaptive event-triggered communication was introduced in [20]. In [21], the author explored both event-triggered and time-triggered algorithms to solve DOP. In [22], the paper examined the prescribed-time optimization problem of MASs under two control protocols, specifically the continuous-time protocol and its event-triggered control protocol.
All of the above results focused on networks without time delay. In fact, the hardware performance of each agent requires a certain input {{time delay}} for effective communication and processing of information. Hence, the analysis and management of time delay is crucial in understanding and improving the performance of MASs, particularly in cases where there are a large number of agents and intricate communication networks. In the works [23,24,25,26], DOP of MASs with time delay have been studied.
Inspired by the previously discussed works, this paper investigates an event-triggered distributed optimization algorithm with time delay under an undirected communication graph. Each agent utilizes its own gradient information as well as delayed information from its neighbors and itself to search for solutions. The connection between the equilibrium point of MASs and optimal solutions are discussed. Compared with the algorithms in [27,28,29], our algorithm does not require continuous communication between agents and it takes into account that there is a delay in the communication between agents. The algorithm proposed here is based on event-triggered transmissions that can alleviate the communication load and decrease the frequency of controller iterates, resulting in saved network resources. Furthermore, the stability property of MASs with time delay is analyzed using the Lyapunov stability theory, and the Zeno-behavior of triggering time sequence is excluded.
The rest of this article is arranged as follows. The basic theories of graph theory and dynamic systems are provided in Section 2. Section 3 presents the main results. In Section 4, a numerical example is given, which illustrates the effectiveness of the proposed algorithm. Concluding remarks are made in Section 5.
Notations: Let R, Rn, and Rn×n be a set of real numbers, n-dimensional real vectors, and n×n real matrices, respectively. Let In∈Rn×n be the identity matrix. 1n=[1,⋯,1]T. AT is the transpose of A. ∇fi(x)=∂fi(x)∂x is gradient of the function fi(x). ⊗ is the Kronecker product. ‖A‖ denotes the induced matrix norm and ‖⋅‖ represents the Euclidean norm for x∈Rn. λmax(A) is the largest eigenvalue of symmetric matrix A.
In this section, the concepts related to graph theory will be introduced, and then the distributed optimization problem is presented.
Let G={V,E,A} be an undirected and connected graph used to model a network of N agents, where V={1,2,...,N} denotes the set of nodes and E={(i,j)∈V×V} denotes the set of edges. Let A=[aij] be the adjacency matrix of graph G, and aij=aji=1 if (i,j)∈E and aij=0, otherwise. Ni={j∈V:(i,j)∈E} is the set of neighbors of node i. The degree matrix of graph G is denoted by D=diag{d1,d2,...,dN} and di=∑j∈Niaij. Furthermore, the Laplacian matrix L=[lij] is defined as L=D−A. If graph G is a connected undirected graph, one can derive that the eigenvalues of matrix L satisfy 0=λ1<λ2≤...≤λN.
The following DOP is studied in this paper
minxi∈RnN∑i=1fi(xi), | (2.1) |
where each agent i∈V has a local convex function fi:Rn→R. Each of the functions fi is private and accessible only to the corresponding agent i. The aim of each agent is to solve the optimization problem in (2.1) cooperatively, and the interaction topology among the agents is expressed by graph G. For convenience, denote f(x)=∑Ni=1fi(xi), x=[xT1,...,xTN]T∈RNn.
Consider the MASs with each agent modeled by first-order dynamics as follows:
˙xi(t)=ui(t),i=1,2,...,N, | (2.2) |
where xi(t)∈Rn represents state vector of agent i, and ui(t)∈Rn is the control input.
The primary objective in this article is to design distributed event-triggered optimization algorithms ui(t) for system (2.2), with the aim of driving the position state of N agents toward the optimal solution x∗ of Eq (2.1). Toward this end, the following definition is proposed.
Definition 1: DOP (2.1) for MAS (2.2) is solved if
limt→∞‖x(t)−x∗‖=0, |
where x∗=argminx∈RNnf(x). Thus, x∗ is considered as the global optimal solution, which minimizes the global objective function.
The following assumptions will be used in this article.
Assumption 1: The communication topology graph G is undirected and connected.
Assumption 2: Each local cost function fi is twice continuously differentiable and ωi-strongly convex (ωi>0) over Rn. Additionally, ▽fi is ℓi-Lipschitz (ℓi>0) for each i∈V.
Remark 1: A function fi is said to be twice continuously differentiable and ωi-strongly convex if ∀K⊂Rn is a convex and compact set and if there is a positive number ω=mini∈Vωi that makes (∇fi(k1)−∇fi(k2))T(k1−k2)≥ω‖k1−k2‖2, ∀k1,k2∈K. ▽fi is ℓi-Lipschitz with positive constant ℓ=maxi∈V{ℓi}, and one has ‖∇fi(k1)−∇fi(k2)‖≤ℓ‖k1−k2‖,∀k1,k2∈K.
This section presents sufficient conditions for DOP (2.1) and excludes Zeno-behavior with the designed event-triggered mechanism for any agent.
To eliminate continuous communication with neighboring agents, the following event-triggered control protocol with time delay is used:
{ui(t)=−N∑j=1aij(xi(tik−℘)−xj(tjk′−℘))−μ∇fi(xi(t))−vi(t)˙vi(t)=N∑j=1aij(xi(tik−℘)−xj(tjk′−℘)), | (3.1) |
where μ, ℘ are positive numbers and vi denotes the auxiliary variable of agent i. The latest triggering instant of agent j is denoted by tjk′=argminl∈Ni,t≥tjl{t−tjl}, and the triggering time sequence is described recursively as follows:
tik+1=inf{t|t>tik,gi(t)≥0} | (3.2) |
and
gi(t)=‖ei(t)‖2−κ1‖N∑j=1aij(xi(tik)−xj(tjk′))‖2−κ2e−γ(t−t0), | (3.3) |
for given constants κ1>0, κ2>0, γ>0. ei(t)=xi(tik)−xi(t) is the state measurement error and ei(t) is equal to 0 at t=tik. ‖∑Nj=1aij(xi(tik)−xj(tjk′))‖ represents the error sum of the states of all agents.
Remark 2: In algorithm (3.1), the controller design is mainly divided into following three parts:
(i) The first part as state feedback for stabilizing system (2.2);
(ii) The second part ∇fi(xi(t)) is gradient direction of the local cost function, which is used to find the optimal solution of the cost function;
(iii) The third part vi(t) is the state auxiliary term, which plays an important role in proving the stability of the algorithm.
With the error ei(t), the control protocol (3.1) can be converted into
{u(t)=−Lx(t−℘)−Le(t−℘)−μ∇˜f(x(t))−v(t)˙v(t)=Lx(t−℘)+Le(t−℘), | (3.4) |
where L=L⊗In∈RNn×Nn, u(t)=[uT1(t),uT2(t),...,uTN(t)]T∈RNn, v(t)=[vT1(t),vT2(t),...,vTN(t)]T∈RNn, e(t−℘)=[eT1(t−℘),eT2(t−℘),...,eTN(t−℘)]T∈RNn and ∇˜f(x(t))=[∇f1(x1(t))T,∇f2(x2(t))T,...,∇fN(xN(t))T]T∈RNn.
Definition 2: If there is an infinite numbers of events in a finite period of time, then the event-triggered time sequence {tik} exhibits Zeno-behavior.
Remark 3: If the event-triggering time sequence exists Zeno-behavior, it implies that there is a positive constant T, such that limk→∞tik=T.
Lemma 1 (Barbalat's lemma [30]): For all t≥t0, the function y:R+→R is uniformly continuous. If
limt→∞∫tt0y(z)dz |
exists and is bounded, then
limt→∞y(t)=0. |
Lemma 2: Suppose Assumptions 1 and 2 hold and ∑Ni=1vi(0)=0n, then (x∗,−μ∇˜f(x∗)) is an equilibrium point of system (3.1), where x∗ is the optimal solution of DOP (2.1).
Proof: By Assumption 1, we have (1N⊗In)TL=0Nn⊗In. One can derive that
(1N⊗In)T˙v(t)=(1N⊗In)T(Lx(t−℘)+Le(t−℘))=0Nn. |
Thus, (1N⊗In)Tv(t)=(1N⊗In)Tv(0)=0Nn. Let (x∗,v∗) be an equilibrium point of system (3.4), then the equilibrium point satisfies
{0Nn=−Lx∗−Le(t−℘)−μ∇˜f(x∗)−v∗0Nn=Lx∗+Le(t−℘). | (3.5) |
By (3.5), we have v∗=−μ∇˜f(x∗) and then (1N⊗In)T∇˜f(x∗)=0Nn, which indicates that x∗ is an optimal solution of DOP (2.1).
For simplicity, let us consider the case with n=1. Notably, the case of n>1 can also be proven using complicated calculations based on the property of the Kronecker product.
Theorem 1: Suppose that Assumptions 1 and 2 hold. Consider DOP (2.1) with the first-order MASs (2.2). The event-triggered control protocol is given by (3.4), where the triggering time sequence for each agent is determined by (3.2). Then, x(t) asymptotically converges to the global minimizer x∗ if there are appropriate positive numbers μ,β∈(0,12(μα1ω−ℓ2)), κ1∈(0,12λmax(LL)), α1>α2>α3>α4 such that the following linear matrix inequalities (LMIs) are feasible:
Λ=[α1000α2IN−1α3IN−10α3IN−1α4IN−1]>0, |
Θ=−[IN−1Θ1Θ2 0 0⋆Θ3α22IN−1Θ4Θ4⋆⋆α3IN−1Θ5Θ5⋆⋆⋆IN−1 0⋆⋆⋆⋆Θ6]<0, |
where Θ1=μ(α2−α1)2IN−1, Θ2=μα32IN−1, Θ3=(ℏ−β)IN−1, Θ4=(α2−α3)2J, Θ5=(α3−α4)2J, Θ6=(β−β℘−2κ1λmax(LTL)1−2κ1λmax(LTL))IN−1. In addition, the Zeno-behavior of the event-triggering time sequence can be excluded.
Proof: Let ˜x=x−x∗, ˜v=v−v∗ and h=∇˜f(˜x+x∗)−∇˜f(x∗). The network dynamics can be rephrased as
{˙˜x(t)=−L˜x(t−℘)−Le(t−℘)−μh(t)−˜v(t)˙˜v(t)=L˜x(t−℘)+Le(t−℘). | (3.6) |
DOP (2.1) is solved by the control protocol (3.1) if limt→∞˜x(t)=0N.
Under Assumption 1, there is an orthogonal matrix Q=(r,R)∈RN×N, such that
QTLQ=[0J], |
where J∈R(N−1)×(N−1) is Jordan matrix and r=1N√N, R∈RN×(N−1) satisfies 1TNR=0N−1,RTR=IN−1.
Let η=QT˜x, ε=QT˜v, δ=QTe. Denote η=(η1,ηT2)T with η1∈R and η2∈RN−1. Similarly, ε=(ε1,εT2)T and δ=(δ1,δT2)T. The systems (3.6) can be rewritten as
{˙η1(t)=−μrTh(t)˙η2(t)=−μRTh(t)−Jη2(t−℘)−ε2(t)−Jδ2(t−℘)˙ε1(t)=0˙ε2(t)=Jη2(t−℘)+Jδ2(t−℘). | (3.7) |
Denote Φ=(η1,ηT2,εT2)T and let V=V1+V2 be the candidate Lyapunov function, where
V1=12ΦTΛΦ, | (3.8a) |
V2=β∫tt−℘ηT2(z)η2(z)dz, | (3.8b) |
with β>0.
The time derivative of (3.8a) along with (3.7) is given by:
˙V1=α1η1(t)˙η1(t)+α2ηT2(t)˙η2(t)+α4εT2(t)˙ε2(t)+α3εT2(t)˙η2(t)+α3ηT2(t)˙ε2(t)=−μα1η1(t)rTh(t)−μα2ηT2(t)RTh(t)−α2ηT2(t)Jη2(t−℘)−α2ηT2(t)ε2(t)−α2ηT2(t)Jδ2(t−℘)+α4εT2(t)Jη2(t−℘)+α4εT2(t)Jδ2(t−℘)−μα3εT2(t)RTh(t)−α3εT2(t)Jη2(t−℘)−α3εT2(t)ε2(t)−α3εT2(t)Jδ2(t−℘)+α3ηT2(t)Jη2(t−℘)+α3ηT2(t)Jδ2(t−℘). | (3.9) |
By Assumption 2, we have
‖RTh(t)‖2≤‖h(t)‖2≤ℓ2˜xT(t)˜x(t), | (3.10a) |
˜xT(t)h(t)=˜xT(t)(∇˜f(˜x(t)+x∗)−∇˜f(x∗))≥ω˜xT(t)˜x(t). | (3.10b) |
Hence, it follows that
ℓ2˜xT(t)˜x(t)−(RTh(t))TRTh(t)≥0. | (3.11) |
Then, considering the first two terms in ˙V1, by (3.10b) and (3.11), one can deduce that
−μα1η1(t)rTh(t)−μα2ηT2(t)RTh(t)=−μα1˜xT(t)h(t)+μ(α1−α2)ηT2(t)RTh(t)≤−μα1ω˜xT(t)˜x(t)+μ(α1−α2)ηT2(t)RTh(t)+ℓ2˜xT(t)˜x(t)−(RTh(t))TRTh(t)=(ℓ2−μα1ω)ηT(t)η(t)+μ(α1−α2)ηT2(t)RTh(t)−(RTh(t))TRTh(t). | (3.12) |
Combining (3.12), inequality (3.9) can be further transformed into
˙V1≤(ℓ2−μα1ω)η1(t)η1(t)+(ℓ2−μα1ω)ηT2(t)η2(t)+μ(α1−α2)ηT2(t)RTh(t)−(RTh(t))TRTh(t)−(α2−α3)ηT2(t)Jη2(t−℘)−α2ηT2(t)ε2(t)−(α2−α3)ηT2(t)Jδ2(t−℘)+(α4−α3)εT2(t)Jη2(t−℘)+(α4−α3)εT2(t)Jδ2(t−℘)−μα3εT2(t)RTh(t)−α3εT2(t)ε2(t). | (3.13) |
Taking the time derivation of (3.8b), it yields
˙V2=βηT2(t)η2(t)−β(1−℘)ηT2(t−℘)η2(t−℘). | (3.14) |
According to the triggering condition (3.3), one can obtain
‖ei(t)‖2<κ1‖N∑j=1aij(xi(tik)−xj(tjk′))‖2+κ2e−γ(t−t0), |
which means
12‖e(t−℘)‖2<κ12˜xT(t−℘)LTL˜x(t−℘)+κ12˜eT(t−℘)LTL˜e(t−℘)+κ1˜xT(t−℘)LTL˜e(t−℘)+N2κ2e−γ(t−℘−t0)≤κ1(˜xT(t−℘)LTL˜x(t−℘)+˜eT(t−℘)LTL˜e(t−℘))+N2κ2e−γ(t−℘−t0), |
then
‖e(t−℘)‖2<ℵ1‖˜x(t−℘)‖2+ℵ2e−γ(t−t0)=ℵ1ηT(t−℘)η(t−℘)+ℵ2e−γ(t−t0), | (3.15) |
where ℵ1=2κ1λmax(LTL)1−2κ1λmax(LTL), ℵ2=Nκ2eγ℘1−2κ1λmax(LTL).
By (3.13)–(3.15), it follows that
˙V=˙V1+˙V2≤(ℓ2−μα1ω)η1(t)η1(t)+(ℓ2−μα1ω)ηT2(t)η2(t)+μ(α1−α2)ηT2(t)RTh(t)−(RTh(t))TRTh(t)−(α2−α3)ηT2(t)Jη2(t−℘)−α2ηT2(t)ε2(t)−(α2−α3)ηT2(t)Jδ2(t−℘)+(α4−α3)εT2(t)Jη2(t−℘)+(α4−α3)εT2(t)Jδ2(t−℘)−μα3εT2(t)RTh(t)−α3εT2(t)ε2(t)+βηT2(t)η2(t)−β(1−℘)ηT2(t−℘)η2(t−℘)−eT(t−℘)e(t−℘)+eT(t−℘)e(t−℘)<−ℏηT(t)η(t)+HTΘH+ℵ2e−γ(t−t0), | (3.16) |
where ℏ=12(μα1ω−ℓ2), H=[(RTh(t))T,ηT2(t),εT2(t),δT2(t−℘),ηT2(t−℘)]T.
According to Eq (3.16), we are able to get that
V(t)<V(t0)+ℵ2∫tt0e−γ(z−t0)dz. | (3.17) |
As a result, limt→∞V(t) is bounded. Thus, (3.16) enforces that
V(∞)−V(t0)<−ℏ∫∞t0ηT(z)η(z)dz+ℵ2∫∞t0e−γ(z−t0)dz=−ℏ∫∞t0˜xT(z)˜x(z)dz+ℵ2γ, |
then
∫∞t0˜xT(z)˜x(z)dz<1ℏ(V(t0)−V(∞)+ℵ2γ). | (3.18) |
By Lemma 1, ˜x(t) asymptotically converges to zero, namely,
limt→∞‖x(t)−x∗‖=0. |
In the following, it will be proven that there is no Zeno-behavior in the triggering time sequence.
Calculate the upper righthand Dini derivative of ‖ei(t)‖ for any t∈[tik,tik+1) and one obtains that
D+‖ei(t)‖≤‖˙ei(t)‖=‖˙xi(t)‖≤‖N∑j=1aij(xi(tik−℘)−xj(tjk′−℘))‖+μ‖∇fi(xi(t))‖+‖vi(t)‖≤‖L˜x(t−℘)+Le(t−℘)‖+μ‖∇fi(xi(t))‖+‖vi(t)‖. | (3.19) |
Invoking (3.15), we can get that
D+‖ei(t)‖≤(1+ℵ121)‖L‖‖˜x(t−℘)‖+ℵ122‖L‖e−γ2(t−t0)+μ‖∇fi(xi(t))‖+‖vi(t)‖. | (3.20) |
Since ei(tik)=0, one can conclude that
‖ei(t)‖≤∫ttik(1+ℵ121)‖L‖‖˜x(z−℘)‖dz+∫ttikℵ122‖L‖e−γ2(tik−t0)dz+μ∫ttik‖∇fi(xi(z))‖dz+∫ttik‖vi(z)‖dz. | (3.21) |
By (3.17), since V(t) is bounded, one can assume that ‖˜x(t)‖≤ρ1 and ‖vi(t)‖≤ρ2. Moreover, combined with Assumption 2, one can assume that ‖∇fi(xi(t))‖≤ρ3, then it holds that
‖ei(tik+1)‖≤((1+ℵ121)‖L‖ρ1+ℵ122‖L‖e−γ2(tik−t0)+μρ3+ρ2)(tik+1−tik). | (3.22) |
The next event will be triggered only if the value of the driving error crosses the zero threshold. It yields that
‖ei(tik+1)‖2=κ1‖N∑j=1aij(xi(tik+1)−xj(tjk′+1))‖2+κ2e−γ(tik+1−t0)≤((1+ℵ121)‖L‖ρ1+ℵ122‖L‖e−γ2(tik−t0)+μρ3+ρ2)2(tik+1−tik)2. | (3.23) |
If the event-triggering time sequence exhibits Zeno-behavior, we have
0=limk→∞(tik+1−tik)≥limk→∞√κ2e−γ(tik+1−t0)(1+ℵ121)‖L‖ρ1+ℵ122‖L‖e−γ2(tik−t0)+μρ3+ρ2=√κ2e−γ(T−t0)(1+ℵ121)‖L‖ρ1+ℵ122‖L‖e−γ2(T−t0)+μρ3+ρ2>0, | (3.24) |
which is a contradiction, and, thus, Zeno-behavior does not exhibit in the triggering time sequence.
In this section, the effectiveness of the obtained results is illustrated by a simulation example.
Consider the first-order multi-agent systems with six agents, and the topology of communication graph is depicted in Figure 1.
The strongly convex local objective functions are defined as fi(xi)=0.2(xi−i)2+i for i=1,...,6. For any initial value xi(0) and the initial value vi(0), it satisfies ∑61vi(0)=0. Undoubtedly, the assumptions given in this paper are satisfied. Set κ2=0.5, β=0.006, γ=0.5, μ=0.4, ℘=0.04, α1=3.5, α2=1.205, α3=1.203 and α4=1.202. By calculation, one has J=diag{1,2,3,3,5}, Θ1=−0.459I5, Θ2=0.2406I5, Θ3=0.799I5, Θ4=0.001J, Θ5=0.0005J, Θ6=0.0055I5. The range of parameter κ1 can be determined that κ1∈(0,0.02). Let κ1=0.01. The simulation results are displayed in Figures 2–7. Figure 2 depicts the state evolution trajectory of each agent. Figure 3 describes the trajectories of the event-triggered input ui(t). The global objective function converges to an optimal solution 24.5, which is exhibited in Figure 4. Figure 5 illustrates the event-triggering instants for each agent. The minimum intervals between successive triggering events for agent 1 through agent 6 are as follows: 0.135, 0.435, 0.38, 0.4, 0.425 and 0.165. It is worth noting that all these intervals are greater than the sampling step of 0.005s. Therefore, Zeno-behavior does not exhibit in the triggering time sequence. The gradient sum of fi(x) is depicted in Figure 6. The error norm of agent 1 is presented in Figure 7.
This article investigated a distributed optimization problem with the first-order MASs. A distributed event-triggered algorithm that allows the agents of control systems to achieve the optimal trajectory in the case of time delay was designed. The effectiveness of the algorithm was rigorously proved by using the Lyapunov stability theory. The adaptive distributed optimization problem through event-triggered communication will be discussed in future study.
The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.
This work is jointly sponsored by National Natural Science Foundation of China under Grant 61673080, Science and Technology Research Program of Chongqing Municipal Education Commission under Grant KJZD-K202000601, Chongqing Talent Plan under Grant cstc2021ycjh-bgzxm0044.
The authors declare there is no conflict of interest.
[1] |
J. Wang, Y. Yan, Z. Liu, C. L. Chen, C. Zhang, K. Chen, Finite-time consensus control for multi-agent systems with full-state constraints and actuator failures, Neural Networks, 157 (2023), 350–363. https://doi.org/10.1016/j.neunet.2022.10.028 doi: 10.1016/j.neunet.2022.10.028
![]() |
[2] |
Q. Wang, Z. Duan, J. Wang, Q. Wang, G. Chen, An accelerated algorithm for linear quadratic optimal consensus of heterogeneous multi-agent systems, IEEE Trans. Autom. Control, 67 (2022), 421–428. https://doi.org/10.1109/TAC.2021.3056363 doi: 10.1109/TAC.2021.3056363
![]() |
[3] |
Q. Wang, X. Dong, J. Yu, J. L¨u, Z. Ren, Predefined finite-time output containment of nonlinear multi-agent systems with leaders of unknown inputs, IEEE Trans. Circuits Syst. I Regul. Pap., 68 (2021), 3436–3448. https://doi.org/10.1109/TCSI.2021.3083612 doi: 10.1109/TCSI.2021.3083612
![]() |
[4] |
A. Nandanwar, N. K. Dhar, D. Malyshev, L. Rybak, L. Behera, Stochastic event-based super-twisting formation control for multiagent system under network uncertainties, IEEE Trans. Control Network Syst., 9 (2022), 966–978. https://doi.org/10.1109/TCNS.2021.3089142 doi: 10.1109/TCNS.2021.3089142
![]() |
[5] |
Q. Lu, X. Liao, S. Deng, H. Li, Asynchronous algorithms for decentralized resource allocation over directed networks, IEEE Trans. Parallel Distrib. Syst., 34 (2023), 16–32. https://doi.org/10.1109/TPDS.2022.3212424 doi: 10.1109/TPDS.2022.3212424
![]() |
[6] |
X. Shi, Z. Meng, S. Dong, X. Wang, Distributed resource allocation algorithm for second-order multi-agent systems with external disturbances, Int. J. Control, 96 (2022), 2181–2189. https://doi.org/10.1080/00207179.2022.2086927 doi: 10.1080/00207179.2022.2086927
![]() |
[7] | B. Polyak, Introduction to Optimization, Chapman Hall, 1987. |
[8] |
R. Xin, S. Pu, A. Nedic, U. A. Khan, A general framework for decentralized optimization with first-order methods, Proc. IEEE, 108 (2020), 1869–1889. https://doi.org/10.1109/JPROC.2020.3024266 doi: 10.1109/JPROC.2020.3024266
![]() |
[9] | O. Shorinwa, R. N. Haksar, P. Washington, M. Schwager, Distributed multirobot task assignment via consensus ADMM, IEEE Trans. Rob., 39 (2023), 1781–1800. https://doi.org/10.1109/TRO.2022.3228132 |
[10] |
I. Jeong, A review of decentralized optimization focused on information flows of decomposition algorithms, Comput. Oper. Res., 153 (2023), 106190. https://doi.org/10.1016/j.cor.2023.106190 doi: 10.1016/j.cor.2023.106190
![]() |
[11] |
Z. Yu, J. Sun, S.Yu, H. Jiang, Fixed-time distributed optimization for multi-agent systems with external disturbances over directed networks, Int. J. Robust Nonlinear Control, 33 (2023), 953–972. https://doi.org/10.1002/rnc.6408 doi: 10.1002/rnc.6408
![]() |
[12] |
A. Nedic, A. Olshevsky, Stochastic gradient-push for strongly convex functions on time-varying directed graphs, IEEE Trans. Autom. Control, 61 (2016), 3936–3947. https://doi.org/10.1109/TAC.2016.2529285 doi: 10.1109/TAC.2016.2529285
![]() |
[13] |
Y. Tian, Y. Sun, G. Scutari, Achieving linear convergence in distributed asynchronous multiagent optimization, IEEE Trans. Autom. Control, 65 (2020), 5264–5279. https://doi.org/10.1109/TAC.2020.2977940 doi: 10.1109/TAC.2020.2977940
![]() |
[14] |
P. Lin, W. Ren, J. A. Farrell, Distributed continuous-time optimization: nonuniform gradient gains, finite-time convergence, and convex constraint set, IEEE Trans. Autom. Control, 62 (2017), 2239–2253. https://doi.org/10.1109/TAC.2016.2604324 doi: 10.1109/TAC.2016.2604324
![]() |
[15] |
S. Rahili, W. Ren, Distributed continuous-time convex optimization with time-varying cost functions, IEEE Trans. Autom. Control, 62 (2017), 1590–1605. https://doi.org/10.1109/TAC.2016.2593899 doi: 10.1109/TAC.2016.2593899
![]() |
[16] |
W. Zhu, H. Tian, Distributed convex optimization via proportional-integral-differential algorithm, Meas. Control, 55 (2022), 13–20. https://doi.org/10.1177/00202940211029332 doi: 10.1177/00202940211029332
![]() |
[17] |
S. Li, X. Nian, Z. Deng, Distributed optimization of second-order nonlinear multiagent systems with event-triggered communication, IEEE Trans. Control Network Syst., 8 (2021), 1954–1963. https://doi.org/10.1109/TCNS.2021.3092832 doi: 10.1109/TCNS.2021.3092832
![]() |
[18] |
H. Dai, X. Fang, W. Chen, Distributed event-triggered algorithms for a class of convex optimization problems over directed networks, Automatica, 122 (2020), 109256. https://doi.org/10.1016/j.automatica.2020.109256 doi: 10.1016/j.automatica.2020.109256
![]() |
[19] |
T. Adachi, N. Hayashi, S. Takai, Distributed gradient descent method with edge-based event-driven communication for non-convex optimization, IET Control Theory Appl., 15 (2021), 1588–1598. https://doi.org/10.1049/cth2.12127 doi: 10.1049/cth2.12127
![]() |
[20] |
Z. Wu, Z. Li, Z. Ding, Z. Li, Distributed continuous-time optimization with scalable adaptive event-based mechanisms, IEEE Trans. Syst. Man Cybern. Syst., 50 (2020), 3252–3257. https://doi.org/10.1109/TSMC.2018.2867175 doi: 10.1109/TSMC.2018.2867175
![]() |
[21] |
N. Tran, Y. Wang, X. Liu, Distributed optimization problem for second-order multi-agent systems with event-triggered and time-triggered communication, J. Franklin Inst., 356 (2019), 10196–10215. https://doi.org/10.1016/j.jfranklin.2018.02.009 doi: 10.1016/j.jfranklin.2018.02.009
![]() |
[22] |
S. Chen, H. Jiang, Z. Yu, F. Zhao, Distributed optimization of single-integrator systems with prescribed-time convergence, IEEE Syst. J., 17 (2023), 3235–3245. https://doi.org/10.1109/JSYST.2022.3227024 doi: 10.1109/JSYST.2022.3227024
![]() |
[23] |
D. Wang, J. Liu, J. Lian, Y. Liu, Z. Wang, W. Wang, Distributed delayed dual averaging for distributed optimization over time-varying digraphs, Automatica, 150 (2023), 110869. https://doi.org/10.1016/j.automatica.2023.110869 doi: 10.1016/j.automatica.2023.110869
![]() |
[24] |
X. Xu, Z. Yu, D. Huang, H. Jiang, Distributed optimization for multi-agent systems with communication delays and external disturbances under a directed network, Nonlinear Anal.-Model. Control, 28 (2023), 1–19. https://doi.org/10.15388/namc.2023.28.31563 doi: 10.15388/namc.2023.28.31563
![]() |
[25] |
J. Yan, H. Yu, X. Xia, Distributed optimization of multi-agent systems with delayed sampled-data, Neurocomputing, 296 (2018), 100–108. https://doi.org/10.1016/j.neucom.2018.03.036 doi: 10.1016/j.neucom.2018.03.036
![]() |
[26] |
C. Wang, S. Xu, D. Yuan, Cooperative convex optimization with subgradient delays using push-sum distributed dual averaging, J. Franklin Inst., 358 (2021), 7254–7269. https://doi.org/10.1016/j.jfranklin.2021.07.015 doi: 10.1016/j.jfranklin.2021.07.015
![]() |
[27] | Z. Deng, J. Luo, Fully distributed algorithms for constrained nonsmooth optimization problems of general linear multi-agent systems and their application, IEEE Trans. Autom. Control, 2023. https://doi.org/10.1109/TAC.2023.3301957 |
[28] |
H. Zhou, X. Zeng, Y. Hong, Adaptive exact penalty design for constrained distributed optimization, IEEE Trans. Autom. Control, 64 (2019), 4661–4667. https://doi.org/10.1109/TAC.2019.2902612 doi: 10.1109/TAC.2019.2902612
![]() |
[29] |
Z. Deng, T. Chen, Distributed algorithm design for constrained resource allocation problems with high-order multi-agent systems, Automatica, 14 (2022), 110492. https://doi.org/10.1016/j.automatica.2022.110492 doi: 10.1016/j.automatica.2022.110492
![]() |
[30] | P. A. Ioannou, J. Sun, Robust Adaptive Control, Upper Saddle River, NJ, USA: Prentice-Hall, 1995. |
1. | Siman Lin, Xiaotang Zhang, Junwu Ren, Manchun Tan, Zero gradient sum algorithm of arbitrary initial value with constraints and communication delay based on directed graph, 2024, 0020-7721, 1, 10.1080/00207721.2024.2392829 | |
2. | Wentai Wu, Chun Zhang, Tiancheng Hu, 2024, Fast Consistency of Second-order Multi-agent Systems based on Two-layer Neighbor Algorithm, 979-8-3503-8778-0, 1562, 10.1109/CCDC62350.2024.10588212 | |
3. | He Bai, Kinjal Bhar, Jemin George, Carl E. Busart, 2024, 9780128035818, 10.1016/B978-0-443-14081-5.00038-6 |