| 引用本文: | 齐美彬， 程佩琳， 靳学明， 张什永， 项厚宏. 基于密集连接卷积网络的雷达辐射源信号分选[J]. 雷达科学与技术, 2022, 20(6): 635-642.[点击复制] |
| QI Meibin, CHENG Peilin, JIN Xueming, ZHANG Shiyong, XIANG Houhong. Radar Emitter Signal Sorting Based on Densely Connected Convolutional Networks[J]. Radar Science and Technology, 2022, 20(6): 635-642.[点击复制] |
|
| 摘要: |
| 针对现代战场电磁环境下复杂调制雷达信号分选准确率低的问题，本文提出一种基于密集连接卷积网络(Densely Connected Convolutional Networks, DenseNet)的雷达辐射源信号分选算法。该算法采用脉冲描述字(Pulse Description Word, PDW)参数与脉内参数相结合作为分选特征，并对特征参数进行差值预处理生成训练数据，使用一维DenseNet网络进行分选。采用本文预处理方法可以有效提取特征间的相关性差异，同时弥补脉间参数PDW对脉内调制信息的缺失。实验结果表明，该方法能有效实现复杂雷达辐射源信号的分选，在0 dB的信噪比下可以达到98%以上的分选准确率。 |
| 关键词: 雷达信号分选 脉间特征 脉内特征 密集神经网络 |
| DOI：DOI:10.3969/j.issn.1672-2337.2022.06.006 |
| 分类号:TN971 |
| 基金项目: |
|
| Radar Emitter Signal Sorting Based on Densely Connected Convolutional Networks |
|
QI Meibin, CHENG Peilin, JIN Xueming, ZHANG Shiyong, XIANG Houhong
|
|
1. School of Computer and Information Engineering, Hefei University of Technology, Hefei 230009, China;2. The 38th Research Institute of China Electronics Technology Group Corporation, Hefei 230088, China
|
| Abstract: |
| To address the problem of low accuracy in complex modulated radar signal sorting in modern battlefield electromagnetic environments, this paper proposes a radar emitter signal sorting algorithm based on densely connected convolutional networks (DenseNet). The algorithm uses the combination of pulse description word (PDW) parameters and intra-pulse parameters as the sorting features, preprocesses the feature parameters to generate training data, and uses one-dimensional DenseNet for sorting. The proposed preprocessing method can extract correlation differences between features and compensate for the lack of intra-pulse modulation information caused by the inter-pulse parameter PDW. The experimental results show that this method can effectively realize the sorting of complex radar emitter signals, and the sorting accuracy can reach more than 98% under the signal-to-noise ratio of 0 dB. |
| Key words: radar signal sorting inter-pulse feature intra-pulse feature DenseNet |
