面向双机编队战术识别的时空图卷积网络  

Spatiotemporal Graph Convolutional Network for Tactical Recognition of Dual-aircraft Formation

作　　者：李岩山[1] 苏伟鹏 贺嘉璠 陈佳馨 LI Yanshan;SU Weipeng;HE Jiafan;CHEN Jiaxin(College of Electronics and Information Engineering,Shenzhen University,Shenzhen 518000,China;Science and Technology on Information Systems Engineering Laboratory,Nanjing 210023,China)

机构地区：[1]深圳大学电子与信息工程学院,深圳518000 [2]信息系统工程重点实验室,南京210023

出　　处：《指挥信息系统与技术》2023年第3期25-30,共6页Command Information System and Technology

基　　金：信息系统工程重点实验室开放基金(05202206)资助项目。

摘　　要：空战中的双机编队战术识别具有重要的研究意义和应用价值。为了提高双机编队战术识别的准确率和鲁棒性,采用深度学习技术,提出了一种面向双机编队战术识别的时空图卷积网络。首先,对双机编队协同作战序列数据进行了时空图构建;然后,通过空域图卷积网络模块、时域图卷积网络模块和通道注意力模块提取了双机编队战术的时空域特征;最后,利用分类器对双机编队战术进行了识别。试验结果表明,该网络可有效识别双机编队战术。The tactical recognition of dual-aircraft formation in air combat has important research sig⁃nificance and application value.In order to improve the accuracy and robustness of tactical recognition of dual-aircraft formation,a spatiotemporal graph convolutional network for tactical recognition of du⁃al-aircraft formation is proposed by using the deep learning technology.Firstly,the spatiotemporal graph is constructed for the sequence data of dual-aircraft formation cooperative combat.Then,the spatiotemporal domain features of dual-aircraft formation tactics are extracted through the graph convo⁃lutional network(GCN)module,the temporal convolutional network(TCN)module,and the chan⁃nel attention module.Finally,the classifier is used to recognize the dual-aircraft formation tactics.Ex⁃perimental results show that the network can effectively recognize the tactics of dual-aircraft formation.

关 键 词：双机编队战术 时空图卷积网络 注意力机制 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]