具有非协同步长的噪声鲁棒分布式资源分配算法  

Noise-robust distributed resource allocation with uncoordinated step-sizes

作　　者：吴温文 朱善迎 刘帅[3] 关新平[1,2] Wenwen WU;Shanying ZHU;Shuai LIU;Xinping GUAN(Department of Automation,Shanghai Jiao Tong University,Shanghai 200240,China;Key Laboratory of System Control and Information Processing,Ministry of Education,Shanghai 200240,China;School of Control Science and Engineering,Shandong University,Jinan 250100,China)

机构地区：[1]上海交通大学自动化系,上海200240 [2]系统控制与信息处理教育部重点实验室,上海200240 [3]山东大学控制科学与工程学院,济南250100

出　　处：《中国科学：信息科学》2024年第3期692-707,共16页Scientia Sinica(Informationis)

基　　金：国家自然科学基金(批准号:62173225,62133008,61821004)资助项目。

摘　　要：针对通信噪声干扰下的资源分配问题,本文提出了基于对偶梯度追踪的分布式算法以实现资源的最优配置.通过设计新的噪声追踪策略与引入抑制系数,本文算法对噪声干扰具有更好的鲁棒性,并且支持各节点使用非协同步长进行迭代更新.理论分析证明了对于非Lipschitz光滑甚至不可微的强凸目标函数,算法在噪声干扰下仍能收敛到最优解邻域并且具有O(1=k)收敛速率.最后,在IEEE 14母线和IEEE 118母线系统上验证了本文所提出分布式算法的有效性和可拓展性.To tackle the resource allocation problem with noisy information exchange,a noise-robust distributed dual gradient tracking algorithm is proposed.By integrating the novel designed noise tracing strategy and suppression parameters into the dual gradient tracking protocol,the proposed algorithm obtains better robustness against noise interference and it can operate with uncoordinated step-sizes.For strongly convex,non-Lipschitz smooth or even non-differentiable cost functions,the convergence of the algorithm can be established under standard assumptions and it has an O(1=k)convergence rate.Finally,the algorithm is tested on IEEE 14-bus and IEEE 118-bus systems to demonstrate its effectiveness and scalability.

关 键 词：分布式优化 对偶问题 噪声抑制 资源分配 非协同步长 

分 类 号：TM73[电气工程—电力系统及自动化]