Distributed optimization of first-order discrete-time multi-agent systems with event-triggered communication
Distributed optimization of first-order discrete-time multi-agent systems with event-triggered communication
复制标题
DOI：
10.1016/j.neucom.2017.01.021
复制
发表时间：
2017-04
期刊：
影响因子：
6
通讯作者：
Qingguo Lü;Huaqing Li;Dawen Xia
中科院分区：
文献类型：
--
作者：
Qingguo Lü;Huaqing Li;Dawen Xia
研究了一类基于一阶离散多智能体系统的无向网络凸优化问题的事件触发分布式子梯度算法。整个网络的通信过程由一组触发条件控制，由每个agent监控。每个智能体的触发条件和事件触发分布式子梯度优化算法是完全去中心化的，仅取决于每个智能体及其相邻智能体在自身事件触发序列中的个体状态以及每个智能体的局部目标函数。在每个时间瞬间，每个智能体通过使用自己的目标函数和在各自事件触发的时间瞬间从自身及其相邻智能体收集的状态来更新其状态。在无向网络拓扑连通和设计参数设计合理的情况下，建立了保证一致性和得到最优解的充分条件。理论分析表明，事件触发分布式次梯度算法能够使整个智能体网络渐近收敛到凸优化问题的最优解。仿真结果验证了算法的有效性，验证了理论分析的可行性。
This paper focuses on the event-triggered distributed subgradient algorithms for solving a class of convex optimization problems based on first-order discrete-time multi-agent systems over undirected networks. The communication process of the whole network is controlled by a set of trigger conditions monitored by each agent. The trigger condition and event-triggered distributed subgradient optimization algorithm for each agent are completely decentralized and just rest with each agent's and its neighboring agents’ individual states at the event-triggered sequence of themselves as well as each agent's local objective function. At each time instant, each agent updates its state by employing its own objective function and the states collected from itself and its neighboring agents at their separate event-triggered time instants. A sufficient condition for ensuring the consensus and reaching the optimization solution is established under the condition that the undirected network topology is connected and the design parameters are properly designed. Theoretical analysis shows that the event-triggered distributed subgradient algorithm is capable of steering the whole network of agents asymptotically converge to an optimal solution of the convex optimization problem. Simulation results validate effectiveness of the introduced algorithm and demonstrate feasibility of the theoretical analysis.