arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2504.04475eess.SYcs.SY
收藏

多 Euler-Lagrange 系统联盟博弈中的 Nash 均衡寻求:分析及其在 USV 集群对抗中的应用

Constrained coalition games for USV swarm confrontation

School of Automation, Southeast University(东南大学自动化学院);School of Interdisciplinary Science, Beijing Institute of Technology(北京理工大学交叉科学研究院);School of Data Science, Lingnan University(岭南大学数据科学学院);Faculty of Computer Science and Control Engineering, Shenzhen University of Advanced Technology(深圳先进技术大学计算机科学与控制工程学院)

Cheng Yuwen, Jialing Zhou, Meng Luan, Guanghui Wen, Tingwen Huang

AI总结:

针对受未知扰动且含局部与耦合约束的多 EL 系统联盟博弈,提出融合自适应、符号函数、原始-对偶和一致性的分布式 NE 寻求策略,并通过 USV 集群对抗仿真验证。

摘要:

本文研究一类联盟博弈中的 Nash 均衡(NE)寻求问题,该问题同时包含局部约束和耦合约束,并涉及受到未知界扰动的多个 Euler-Lagrange(EL)系统。在每个联盟内部,智能体合作最小化一个共享代价函数,同时与其他联盟展开竞争。文中提出一种在信息约束下寻求 NE 的分布式策略,其中每个智能体只能获得自身的动作、代价函数和约束参数。在所提出的分布式 NE 寻求策略中,自适应技术与符号函数相结合,用于处理 EL 系统中的模型不确定性和未知界扰动。为处理与局部约束和耦合约束相关的 Lagrange 乘子,原始-对偶技术与一致性协议相结合。此外,采用动态平均一致性算法估计联盟代价函数的梯度,并利用领导者跟随协议估计其他智能体的动作。在标准凸性和图连通性假设下,建立了闭环 EL 系统全局收敛至 NE 的结论。作为说明性应用,将涉及编队、合围和拦截任务的无人水面艇(unmanned surface vehicle,USV)集群对抗建模于联盟博弈框架内,并在该模型下开展数值仿真,以验证理论结果。

英文摘要:

This paper formulates unmanned surface vehicle (USV) swarm confrontation as a constrained coalition game, capturing cooperation within swarms and competition between swarms. Local task costs are aggregated into coalition objectives to account for tradeoffs among target approach, interception, and formation maintenance, subject to individual geographic constraints and supply line limits shared within each swarm. For prescribed formation sizes and task assignments, explicit linear matrix inequalities in the task weights are derived to establish strong monotonicity of the coalition pseudogradient, thereby guaranteeing the existence of a unique Nash equilibrium (NE) whenever the joint feasible set is nonempty. A distributed robust NE seeking strategy is then developed by integrating gradient tracking and action estimation with projected primal-dual dynamics using local information and neighbor communication. Adaptive compensation addresses unknown hydrodynamic parameters and bounded time-varying disturbances. The disturbance compensation gains are adjusted through local error feedback without requiring disturbance measurements, prior knowledge of disturbance bounds, or a prescribed disturbance generation model. Under the derived weight conditions, the stated dynamics and communication assumptions, and suitable gain conditions, the physical states of all USVs are shown to converge asymptotically to the unique feasible NE. Numerical simulations demonstrate the effectiveness of the proposed algorithm.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑