系统工程与电子技术 ›› 2024, Vol. 46 ›› Issue (10): 3356-3364.doi: 10.12305/j.issn.1001-506X.2024.10.13

• 传感器与信号处理 • 上一篇    

基于改进残差神经网络的雷达信号识别方法

聂千祁, 沙明辉, 朱应申   

  1. 北京无线电测量研究所, 北京 100854
  • 收稿日期:2023-11-03 出版日期:2024-09-25 发布日期:2024-10-22
  • 通讯作者: 沙明辉
  • 作者简介:聂千祁 (1997—), 男, 硕士研究生, 主要研究方向为电子对抗
    沙明辉 (1986—), 男, 研究员, 博士, 主要研究方向为电子对抗、雷达抗干扰技术
    朱应申 (1985—), 男, 高级工程师, 硕士, 主要研究方向为电子对抗、技术侦察

Radar signal recognition method based on improved residual neural network

Qianqi NIE, Minghui SHA, Yingshen ZHU   

  1. Beijing Institute of Radio Measurement, Beijing 100854, China
  • Received:2023-11-03 Online:2024-09-25 Published:2024-10-22
  • Contact: Minghui SHA

摘要:

针对低信噪比情况下, 雷达信号特征提取困难, 导致识别准确率较低的问题, 提出一种基于改进残差神经网络的雷达信号调制识别方法。首先使用时频分析方法, 将时域信号转化为二维时频图像; 然后对图像进行灰度化、高斯滤波、双线性插值、归一化等预处理, 作为深度学习模型的输入; 最后搭建改进的残差神经网络, 利用空间和通道重构单元减少特征冗余, 提高特征提取精度, 从而提高低信噪比下雷达信号识别准确率。仿真实验结果表明, 信噪比为-8 dB时, 该方法对12类典型雷达信号的整体识别准确率达到96.67%, 具有较好的噪声鲁棒性与抗混淆能力。

关键词: 辐射源信号识别, 时频分析, 深度学习, 残差神经网络

Abstract:

In the case of low signal to noise ratio, radar signal feature extraction is difficult, resulting in low recognition accuracy, a radar signal modulation recognition method based on improved residual neural network is proposed. Firstly, the time-frequency analysis method is used to transform the time-domain signal into a two-dimensional time-frequency image. Then, the image is preprocessed by graying, Gaussian filtering, bilinear interpolation, normalization, etc., as the input of the deep learning model. Finally, an improved residual neural network is built, which uses space and channel reconstruction units to reduce feature redundancy and improve feature extraction accuracy, thereby improving radar signal recognition accuracy under low signal to noise ratio. Simulation results show that when the signal to noise ratio is -8 dB, the overall recognition accuracy of the proposed method for 12 typical radar signals reaches 96.67%, which has good noise robustness and anti-confusion ability.

Key words: emitter signal recognition, time-frequency analysis, deep learning, residual neural network

中图分类号: