基于迁移深度学习的雷达信号分选识别
Radar Signal Sorting and Recognition Based on Transferred Deep Learning
DOI: 10.12677/CSA.2019.99198, PDF,  被引量    国家自然科学基金支持
作者: 王功明：战略支援部队信息工程大学，河南 郑州;中国人民解放军93986部队，新疆 和田；陈世文*, 黄 洁, 黄东华：战略支援部队信息工程大学，河南 郑州
关键词: 雷达信号；分选识别；时频分析；深度学习；迁移学习；Radar Signal； Sorting and Recognition； Time-Frequency Analysis； Deep Learning； Transfer Learning
摘要: 针对当前雷达信号分选识别算法普遍存在的低信噪比下识别能力差、特征参数提取困难、分类器模型参数复杂等问题，提出了一种基于时频分析、深度学习和迁移学习融合模型的雷达信号自动分选识别算法。首先通过引入的多重同步压缩变换得到雷达信号的时频图像，然后利用灰度化、维纳滤波、双三次插值法和归一化等手段对时频图像进行预处理，最后基于迁移学习的方法，以GoogLeNet和ResNet模型为基础完成了对雷达信号的离线训练和在线识别。仿真结果表明，在信噪比为−6 dB时，该算法对9种雷达信号(CW, LFM, NLFM, BPSK, MPSK, Costas, LFM/BPSK, LFM/FSK, BPSK/FSK)的整体平均识别率可达93.4%，较常规人工提取算法具有更好的抗噪性和泛化能力。
Abstract: Aiming at the problems of poor recognition ability under low signal-to-noise ratio (SNR), difficulty in extracting feature parameters and complexity of classifier model parameters commonly existing in current radar signal sorting and recognition algorithms, an automatic radar signal sorting and recognition algorithm based on time-frequency analysis, deep learning and transfer learning fusion model is proposed. Firstly, the time-frequency image of radar signal is obtained by introducing Mul-tisynchrosqueezing Transform. Then, the time-frequency image is preprocessed by gray scale, Wiener filtering, bicubic interpolation and normalization. Finally, based on the migration learning method, the off-line training and on-line recognition of radar signals are completed on the basis of Goog-LeNet and ResNet models. Simulation results show that when SNR is −6 dB, the overall average recognition rate of the algorithm for nine radar signals (CW, LFM, NLFM, BPSK, MPSK, Costas, LFM/BPSK, LFM/FSK, BPSK/FSK) can reach 93.4%, which is better than the conventionally artificial extraction algorithm in noise resistance and generalization.
文章引用：王功明, 陈世文, 黄洁, 黄东华. 基于迁移深度学习的雷达信号分选识别[J]. 计算机科学与应用, 2019, 9(9): 1761-1778. https://doi.org/10.12677/CSA.2019.99198

参考文献

[1] 毛校洁. 雷达信号脉内调制类型识别方法研究[D]: [硕士学位论文]. 哈尔滨: 哈尔滨工程大学, 2019.
[2] 周志文, 黄高明, 陈海洋, 等. 雷达辐射源识别算法综述[J]. 电讯技术, 2017, 57(8): 973-980.
[3] 曲志昱, 毛校洁, 侯长波. 基于奇异值熵和分形维数的雷达信号识别[J]. 系统工程与电子技术, 2018, 40(2): 303-307.
[4] 普运伟, 金炜东, 朱明, 等. 雷达辐射源信号模糊函数主脊切面特征提取方法[J]. 红外与毫米波学报, 2008, 27(2): 133-137.
[5] 王星, 呙鹏程, 田元荣, 等. 基于BDS-GD的低截获概率雷达信号识别[J]. 北京航空航天大学学报, 2018, 44(3): 583-592.
[6] Dong, X., Cheng, S., Yang, J., et al. (2019) Radar Specific Emitter Recognition Based on DBN Feature Extraction. Journal of Physics: Conference Series, 1176, Article ID: 032025. [Google Scholar] [CrossRef] 
[7] Kong, S.-H., Kim, M., Hoang, L.M., et al. (2018) Automatic LPI Radar Waveform Recognition Using CNN. IEEE Access, 6, 4207-4219. [Google Scholar] [CrossRef] 
[8] Wan, J., Yu, X. and Guo, Q. (2019) LPI Radar Waveform Recognition Based on CNN and TPOT. Symmetry, 11, 725. [Google Scholar] [CrossRef] 
[9] Zhou, Z., Huang, G., Chen, H., et al. (2018) Automatic Radar Waveform Recognition Based on Deep Convolutional Denoising Auto-Encoders. Circuits Systems & Signal Processing, No. 5, 1-15.
[10] 张穆清, 王华力, 倪雪. 基于深度学习与支持向量机的低截获概率雷达信号识别[J]. 科技导报, 2019, 37(4): 69-75.
[11] 周东青, 王玉冰, 王星, 等. 基于深度限制波尔兹曼机的辐射源信号识别[J]. 国防科技大学学报, 2016, 38(6): 136-141.
[12] Qu, Z., Mao, X. and Deng, Z. (2018) Radar Signal Intra-Pulse Modulation Recognition Based on Convolutional Neural Network. IEEE Access, 6, 43874-43884. [Google Scholar] [CrossRef] 
[13] Zhang, M., Diao, M., Gao, L., et al. (2017) Neural Networks for Radar Waveform Recognition. Symmetry, 9, 75. [Google Scholar] [CrossRef] 
[14] Pan, S.J. and Yang, Q. (2010) A Survey on Transfer Learning. IEEE Transactions on Knowledge & Data Engineering, 22, 1345-1359. [Google Scholar] [CrossRef] 
[15] Gao, L., Zhang, X., Gao, J., et al. (2019) Fusion Image Based Radar Signal Feature Extraction and Modulation Recognition. IEEE Access, 7, 13135-13148. [Google Scholar] [CrossRef] 
[16] 刘延姣. 雷达信号的特征信息提取与识别技术研究[D]: [硕士学位论文]. 西安: 西安电子科技大学, 2018.
[17] 王爽. 基于时频分析和模糊函数的LPI雷达波形识别算法研究[D]: [硕士学位论文]. 哈尔滨: 哈尔滨工程大学, 2019.
[18] Yu, G., Wang, Z. and Zhao, P. (2019) Multisynchrosqueezing Transform. IEEE Transactions on Industrial Electronics, 66, 5441-5455. [Google Scholar] [CrossRef] 
[19] Tan, C., Sun, F., Kong, T., et al. (2018) A Survey on Deep Transfer Learning. 27th International Conference on Artificial Neural Networks, Rhodes, 4-7 October 2018, 270-279. [Google Scholar] [CrossRef]