摘要
为解决传统复杂雷达信号特征参数提取困难而无法有效进行个体识别和工作模式识别的问题,提出一种基于图像化特征的雷达信号个体识别与工作模式识别方法.首先,利用变换域特征提取将雷达时域信号映射到图像域;然后,通过协同训练来对雷达信号进行个体识别;最后,在个体识别的基础之上,采用图像化特征重构方法和图像分类技术完成对工作模式的识别.实验结果表明,13类个体识别中,利用ResNet和BiLSTM网络协同训练的识别准确率能到82%以上;4类工作模式识别中,利用ResNet网络的识别准确率能到95%以上.
It is difficult to extract the characteristic parameters of the traditional complex radar signal,and thus the individual recognition and operating mode recognition can not be carried out effectively.In order to solve this problem,the paper proposes a radar signal individual recognition and operating mode recognition method based on image features.Firstly,the radar time domain signal is mapped to the image domain by feature extraction in the transform domain.Then,individual recognition of radar signal is carried out through collaborative training.Finally,on the basis of individual recognition,the image feature reconstruction method and image classification technology are used to complete the recognition of operating mode.The experimental results show that the recognition accuracy of using ResNet and BiLSTM network can reach more than 82%in the 13 classifications of individual recognition,and that in the 4 classifications of radar operating mode recognition,the recognition accuracy of using ResNet network can reach more than 95%.
作者
万海东
蔡伟
刘则林
WAN Haidong;CAI Wei;LIU Zelin(No.51 Research Institute,China Electronics Technology Group Corporation,Shanghai 201802,China)
出处
《空天预警研究学报》
2022年第2期79-84,共6页
JOURNAL OF AIR & SPACE EARLY WARNING RESEARCH
关键词
变换域特征提取
协同训练
个体识别
特征重构
工作模式识别
feature extraction in transform domain
collaborative training
individual recognition
feature reconstruction
operating mode recognition
