非平衡有向网络的完全分布式凸优化

分布式凸优化问题的目的是如何以分布式方法最小化局部智能体成本函数和,而现有分布式算法的控制步长选取依赖于系统智能体个数、伴随矩阵等全局性信息,有悖于分布式算法的初衷.针对此问题,提出一种基于非平衡有向网络的完全分布式凸优化算法(FDCOA).基于多智能体一致性理论和梯度跟踪技术,设计了一种非负余量迭代策略,使得FDCOA的控制步长收敛范围仅与智能体局部信息相关,进而实现控制步长的分布式设置.进一步分析了FDCOA在固定强连通和时变强连通网络情形下的收敛性.仿真结果表明本文构建的分布式控制步长选取方法对FDCOA在有向非平衡下的分布式凸优化问题是有效的.     

具有通信时延和有向网络的多智能体分布式优化

本文对非平衡有向拓扑下一阶多智能体系统的分布式优化问题进行研究.研究的智能体在与邻居的通信过程中都有一个变化的时延.本文的目标是找到使得目标函数f(x)=(NΣi=1)fi(xi)最小的智能体的状态.提出了一种基于有向图的拉普拉斯矩阵的零特征值对应的左特征向量和智能体的局部信息的控制器.在这项研究中,去掉了fi(xi)...     

带有不连续动力学的非线性多智能体系统固定时间一致性EI北大核心CSCD

本文研究了一类带有不连续动力学和有界扰动的非线性多智能体系统领导跟随固定时间一致性问题.首先,在不对称的有向拓扑图下,本文设计了一种辅助信号,该辅助信号用于观测领导者状态,且该辅助信号不在通信信道中传输,可以有效地减少系统计算代价.随后,基于辅助信号,本文设计了一种不连续控制协议,以实现多智能体系统固定时间收敛.然后,利用非光滑分析、Lyapunov稳定性理论及代数图论等证明系统可在任意初始状态下达到固定时间一致.最后,仿真实例进一步验证了理论结果的有效性.     

有向网络分布式优化的Barzilai-Borwein梯度跟踪方法

本文研究有向网络上的分布式优化问题, 其全局目标函数是网络上所有光滑强凸局部目标函数的平均值.受Barzilai-Borwein步长改善梯度方法表现的启发, 本文提出了一种分布式Barzilai-Borwein梯度跟踪方法. 与文献中使用固定步长的分布式梯度算法不同, 所提出的方法中每个智能体利用其局部梯度信息自动地计算其步长. 通过同时使用行随机和列随机权重矩阵, 该方法避免了由特征向量估计引起的计算和通信. 当目标函数是光滑和强凸函数时, 本文证明了该算法产生的迭代序列可以线性地收敛到最优解. 对分布式逻辑回归问题的仿真结果验证了所提出的算法比使用固定步长的分布式梯度算法表现更好     

基于有向通信拓扑的高阶分布式一致性算法

为了提高有向通信拓扑下分布式一致性算法的收敛速度,提出了一种基于有向通信拓扑的高阶分布式一致性算法。该算法通过有向单跳通信,利用有向二跳邻接节点的前多步信息提高分布式一致性算法的收敛速度。对有向通信拓扑下该算法的收敛性能和收敛速度进行了分析和仿真比较。结果显示,该算法在满足一定条件下能收敛到初始状态的平均值,与其他同样利用二跳邻接节点信息的一致性算法相比,具有通信量小、收敛速度更快的特点,但是能容忍的最大通信延时变小。     

基于有向图的分布式连续时间非光滑耦合约束凸优化分析

研究一类分布式优化问题, 其目标是在满足耦合不等式约束和局部可行集约束的情况下使非光滑全局代价函数值最小. 首先, 对原有的分布式连续时间投影算法进行拓展, 结合线性代数理论分析, 设计一个适用于强连通加权平衡有向通信网络拓扑图的算法. 其次, 在局部代价函数和耦合不等式约束函数是非光滑凸函数的假设条件下, 利用Moreau-Yosida函数正则化使目标函数和约束函数近似光滑可微. 然后, 根据强连通加权平衡有向图的分布式连续时间投影算法构造李雅普诺夫函数, 证明该算法下的平衡解是分布式优化问题最优解, 并对算法进行收敛性分析. 最后, 通过数值仿真验证算法的有效性.     

切换网络分布式次梯度优化算法

研究了切换网络的多个体分布式次梯度优化算法。在有向切换网络是周期强连通的且对应的邻接矩阵是随机的而非双随机的条件下,利用非二次李雅普诺夫函数方法证明了所提多个体分布式次梯度优化算法的收敛性。最后,通过仿真实例验证了所提算法的有效性。     

面向平均区域覆盖的多机器人分布式控制EI北大核心CSCD

随着通信技术、传感技术和控制技术的发展,多机器人系统因其良好的鲁棒性,灵活性和可扩展性,在理论研究和工程应用中展现出广阔的前景.区域覆盖是多机器人系统典型应用之一,目前多采用维诺图分割覆盖区域并使用Lloyd算法控制机器人前往维诺图细胞中心.然而传统Lloyd算法存在不平衡问题,即机器人覆盖区域面积大小不一,这降低了多机器人协作效率.针对平均区域覆盖问题,本文提出了一种改进的Lloyd算法,将维诺图中各细胞面积方差引入Lloyd算法,相应地设计了基于梯度下降法的分布式控制器.本文方法降低了维诺图中各细胞面积的方差,改善了Lloyd算法的平衡性,能够实现整个区域面积更为平均的划分与机器人对该区域的覆盖.数值仿真与无人机实物实验均验证了改进算法的有效性.     

基于分布式神经动态优化的综合能源系统多目标优化调度EI北大核心CSCD

研究了基于神经动态优化的综合能源系统(Integrated energy systems,IES)分布式多目标优化调度问题.首先,将IES元件单元(包含负荷)作为独立的决策主体,联合考量其运行成本和排放成本,并计及多能源设备间的传输损耗,提出了IES多目标优化调度模型,该模型可描述为一类非凸多目标优化问题.其次,针对此类问题的求解,提出了一种基于神经动力学系统的分布式多目标优化算法,该算法基于动态权重的神经网络模型,可以解决不可分离的不等式约束问题.该算法计算负担小,收敛速度快,并且易于硬件实现.仿真结果表明,所提算法能同时协调综合能源系统的经济性和环境性这两个冲突的目标,且获得了整个帕累托前沿,有效降低了综合能源系统的污染物排放量和综合运行成本.     

抗干扰的多智能体系统固定时间分布式优化算法

针对一阶多智能体系统提出一种抗干扰的分布式控制算法,在固定时间内解决具有状态约束和外部扰动存在情况下的多智能体系统凸优化问题.该算法分为两部分:第1部分使得每个智能体在任意初始条件下都能在固定时间内收敛到一致;第2部分在满足状态约束条件的同时,使所有局部目标函数的总和在固定时间内取得最小值.该算法能够在外部有界扰动存在的情况下抑制干扰信号,获得最优解,且收敛时间不受初始状态和外部扰动的影响,可以根据任务需求离线地预分配任务建立时间.利用凸优化和固定时间李雅普诺夫稳定性理论证明算法在有界扰动存在时的固定时间收敛性,最后通过智能电网中经济调度问题的实例验证算法的有效性和优越性.     

A new framework for consensus for discrete-time directed networks of multi-agents with distributed delays

In this article, the distributed consensus problem is considered for discrete-time delayed networks of dynamic agents with fixed topologies, where the networks under investigation are directed and the time delays involved are distributed time delays including a single or multiple time delay(s) as special cases. By using the invariance principle of delay difference systems, a new unified framework is established to deal with the consensus for the discrete-time delayed multi-agent system. It is shown that the addressed discrete-time network with arbitrary distributed time delays reaches consensus provided that it is strongly connected. A numerical example is presented to illustrate the proposed methods.     

Fixed-time distributed optimization for multi-agent systems with external disturbances over directed networks

This article considers the fixed-time distributed optimization problem of multi-agent systems with external disturbances, in which the global optimization objective is a convex combination of local objective functions. To solve this issue, a directed communication network is carefully designed, and an integral sliding mode control protocol is proposed based on the gradient of global objective function first. Moreover, two distributed optimal protocols are designed by using the gradient and the Hessian matrix of local objective function, respectively. By employing Lyapunov stability theory, graph theory, convex analysis, and inequality techniques, we prove that all proposed protocols can make agents achieve consensus and converge accurately to the optimal solution of the considered problem in some fixed-time intervals. Finally, some numerical simulations are given to verify the feasibility of the theoretical results.     

Distributed continuous‐time algorithm for a general nonsmooth monotropic optimization problem

This paper investigates a general monotropic optimization problem for continuous‐time networks, where the global objective function is a sum of local objective functions that are only known to individual agent, and general constraints are taken into account, including local inequality constraints, global equality constraint, and local feasible constraints. In addition, all functions involved in the objective functions and inequality constraints are not necessarily differentiable. To solve the problem, a distributed continuous‐time algorithm is designed using subgradient projections, and it is shown that the proposed algorithm is well defined in the sense that the existence of its solutions can be guaranteed. Furthermore, it is proved that the algorithm converges to an optimal solution for the general monotropic optimization problem. Finally, a simulation example is provided for validating the theoretical result.     

A distributed finite-time optimization algorithm for directed networks of continuous-time agents

This article proposes a new distributed finite-time optimization algorithm for agents under directed graphs. By employing the nonsmooth technique and graph theory, a distributed discontinuous algorithm for continuous-time agents subject to strongly convex local cost functions is first designed with a finite-time distributed estimator, where the gradients of the local cost functions are estimated in finite time. It is shown that for a strongly connected graph and arbitrary initial conditions, the proposed algorithms can achieve consensus, and the systems can converge to the optimal point in finite time. Then, a two-step approach is proposed to achieve finite-time optimization of high-order agents with disturbances under directed graphs. Finally, the validity of the proposed finite-time optimization algorithm is verified by two numerical examples.     

Privacy-preserving distributed projected one-point bandit online optimization over directed graphs

The distributed online optimization (DOO) problem with privacy-preserving properties over multiple agents is considered in this paper, where the network model is built by a strongly connected directed graph. To solve this problem, a stochastic bandit DOO algorithm based on differential privacy is proposed. This algorithm uses row- and column-stochastic matrix as the weighting matrices, the requirement of the double random weighting matrix is released. To handle the unknown objective function, the one-point bandit is used to estimate the true gradient information, and the estimated gradient information is used to update of decision variables. Different from the existing DOO algorithms that ignore privacy issues, this algorithm successfully protects the privacy of nodes through a differential privacy policy. Theoretical results show that the algorithm can not only achieve sublinear regret bounds but also protect the privacy of nodes. Finally, simulation results verify the effectiveness of the algorithm.     

网络化分布式凸优化算法研究进展

分布式优化是指利用网络化多自主体之间的协作来求解的一类优化问题,其在大规模数值计算、机器学习、资源分配、传感器网络等方面具有重要的研究意义和应用价值.自主体之间的协作通常基于代数图来描述,且图的结构对分布式优化算法的设计与性能有显著影响.本文针对凸优化问题,基于平衡图和非平衡图的情形,简要讨论了分布式优化算法的最新研究进展,并对今后的发展趋势和应用进行展望.     

基于梯度估计的多智能体系统有限时间分布式优化

现有多智能体系统分布式优化算法大多具有渐近收敛速度,且要求系统的网络拓扑图为无向图或有向平衡图,在实际应用中具有一定的保守性.本文研究了具有强连通拓扑的多智能体系统有限时间分布式优化问题.首先,基于非光滑分析和Lyapunov稳定性理论设计了一个有限时间分布式梯度估计器.然后,基于该梯度估计器提出了一种适用于强连通有向图的有限时间分布式优化算法,实现了多智能体系统中智能体的状态在有限时间内一致收敛到全局最优状态值.与现有的有限时间分布式优化算法相比,新提出的有限时间优化算法适用于具有强连通拓扑的多智能体系统,放宽了系统对网络拓扑结构的要求.此外,本文基于Nussbaum函数方法对上述优化算法进行了拓展解决了含有未知高频增益符号的多智能体系统分布式优化问题.最后,通过仿真实例对提出的分布式优化算法的有效性进行了验证.