基于双延迟深度确定性策略梯度的集群簇间网络拓扑控制方法

打开文本图片集

中图分类号：TP393；TN929.5 文献标识码：A文章编号：2096-4706（2025）24-0130-08

Abstract： To address the cross-cluster communication interruption problem caused by relay node failures in Unmanned Aerial Vehicle （UAV） swarms，this paper proposes a topology control method based on Multi-Agent Twin Delayed Deep Deterministic Policy Gradient （MATD3）.This method constructs topologyand energyconsumption models，and combines an improved A* algorithm withagreedy strategy toeffciently select newrelay nodes.Furthermore，by designinga formation reconstructionrewardfunction，the swarmautonomouslyforms astable regularpolygonformationtoenhancelocalconnectivity. Simulationresultsshowthat theproposed methodsignifcantlyoutperformsalgorithmssuchas Multi-AgentDeepDeterministic Policy Gradient （MADDPG）and Multi-Agent SoftActor-Critic （MASAC） intask eficiency，formation stabilityand energy consumption，providing an efective solution for adaptive topology control of highlydynamic UAV swarm networks.

Keywords：UAV swarm;network connectivity; relay node; topologycontrol

0 引言

无人机（UnmannedAerialVehicles，UAV）因结构简单、成本低、机动性高及隐蔽性强等优点，受到了广泛关注。（剩余12233字）