火力与指挥控制2025，Vol.50Issue(3)：178-185,8.DOI:10.3969/j.issn.1002-0640.2025.03.024
融合数据去噪及神经网络算法的目标威胁判别方法
Target Threat Discrimination Method Based on Denoising of Fused Data and Neural Network Algorithm
李玉玺 1方子穆 1李正宇 1宋振华 1葛尧1
作者信息
- 1. 西安现代控制技术研究所,西安 710065
- 折叠
摘要
Abstract
As an important part of battlefield information intelligent perception,target threat assess-ment plays an important role in modern network centric combat decision-making.In order to effectively and accurately identify whether the enemy target is a threat,the complex problem of target threat assess-ment is transformed into the problem of threat binary classification.Firstly,the six-dimensional charac-teristics of enemy target,including relative distance,relative speed,attack angle,firepower distribution,target type and tactical intention,are selected as the original data set,and the T-SNE data visualization al-gorithm is used for data preprocessing and denoising.Then,the threat classification models are con-structed based on the logistic regression and neural network algorithm respectively.Regularization and dropout strategy are introduced to optimize the model.Finally,a specific example is designed to obtain the accuracy,precision,recall rate and F1 score of the three different algorithms.The experimental results show that the neural network method with regularization and dropout strategy proposed in this paper is the best.关键词
威胁判别/神经网络/T-SNE/线性回归/数据去噪Key words
threat discrimination/neural network/T-SNE/linear regression/data denoising分类
军事科技引用本文复制引用
李玉玺,方子穆,李正宇,宋振华,葛尧..融合数据去噪及神经网络算法的目标威胁判别方法[J].火力与指挥控制,2025,50(3):178-185,8.
