基于动态图卷积神经网络的装甲威胁评估方法
Armored Target Threat Assessment Method Based on Dynamic Graph Convolutional Neural Network
张成佳1
作者信息
摘要
针对传统装甲目标威胁评估方法主观因素影响大、孤立考虑目标特征、忽略时序性的问题,提出一种基于动态图卷积神经网络(Dynamic Graph Convolutional Network,DCGN)的装甲目标威胁评估方法.分析陆域装甲装备作战问题,使用合理的目标特征预处理;根据选取的目标特征,建立威胁评估的初始图;利用动态图卷积神经网络将建立的初始图进行深度学习;最后进行仿真实验,并将评估结果与传统方法进行对比,验证了基于动态图卷积神经网络的装甲目标威胁评估方法在动态不确定、关联性复杂的陆域作战中有更高的准确率、更强的鲁棒性.
Abstract
A threat assessment method for armored targets based on dynamic graph convolutional neural network is proposed to address the issues of large subjective factors,isolated consideration of target features,and neglect of timing in traditional threat assessment methods.This paper analyzes the operational issues of armored equipment on land battlefields,selects reasonable target features to preprocess them,establishes an initial threat assessment graph based on the selected target features,uses dynamic graph convolutional neural networks to perform deep learning on the established initial graph,and conducts a simulation using a numerical example.The evaluation results compared with traditional methods indicate that threat assessment based on dynamic graph convolu-tional neural networks has high accuracy and strong robustness,and is more suitable for practical land battlefield environments with high dynamic and strong correlation.
关键词
威胁评估/陆域装甲装备/动态图卷积神经网络Key words
threat assessment/armored on land battlefield/dynamic graph convolutional neural network引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据