Graph Feature Learning-Based Sea Clutter Suppression Method

doi:10.12382/bgxb.2021.0662

Abstract

Abstract:

In order to reduce the effect of sea clutter on Marine radar,a sea clutter suppression algorithm based on graph feature learning is proposed. The time-frequency transform is used to amplify the dimension of radar echo signal,and based on the idea that the graph structure features can be deeply mined by graph embedding processing, a method of constructing signal node feature vector by graph embedding is presented according to the different structural characteristics of sea clutter and target echo signal in time spectrum. Different from traditional methods such as time domain cancellation and subspace decomposition,this method can be used to implement sea clutter suppression through node classification of different signals in the time spectrum. Simulation and experimental results show that the algorithm can effectively suppress the sea clutter component of radar echo signal,improve the signal to clutter ratio of radar echo signal,and provide a new idea and way for ocean radar to suppress sea clutter.

Key words: sea clutter suppression, time-frequency transformation, graph embedding, graph structure, node classification

CLC Number:

TN957

CHEN Peng, XU Zhen, CAO Zhenxin, WANG Zongxin. Graph Feature Learning-Based Sea Clutter Suppression Method[J]. Acta Armamentarii, 2023, 44(2): 534-544.

Figures/Tables 16

References 25

[1]	任红霞. 雷达海杂波统计建模与仿真[D]. 青岛: 中国海洋大学, 2015.
	REN H X. Statistical modeling and simulation of radar sea clutter[D]. Qingdao: Ocean University of China, 2015. (in Chinese)
[2]	丁鹭飞, 耿富录. 雷达原理[M]. 西安: 西安电子科技大学出版社, 2002.
	DING L F, GENG F L. Radar principle[M]. Xi'an: Xidian University Press, 2002. (in Chinese)
[3]	丁昊, 董云龙, 刘宁波, 等. 海杂波特性认知研究进展与展望[J]. 雷达学报, 2016, 5(5):499-516.
	DING H, DONG Y L, LIU N B, et al. Research progress and prospects of the cognition of sea zooplankton characteristics[J]. Journal of Radar, 2016, 5(5):499-516. (in Chinese)
[4]	姜星宇, 刘宁波, 丁昊, 等. 雷达海杂波时频谱特性分析[J]. 海军航空工程学院学报, 2019, 34(5):413-417.
	JIANG X Y, LIU N B, DING H, et al. Analysis of time-spectrum characteristics of radar sea clutter[J]. Journal of Naval Aeronautical and Astronautical Engineering, 2019, 34(5):413-417. (in Chinese)
[5]	胡昌华, 夏启兵, 周涛, 等. 基于MATLAB的系统分析与设计-时频分析[M]. 西安: 西安电子科技大学出版社, 2002.
	HU C H, XIA Q B, ZHOU T, et al. System analysis and design based on Matlab-time-frequency analysis[M]. Xi'an: Xidian University Press, 2002. (in Chinese)
[6]	马丽文, 张金鹏, 吴家骥, 等. 基于门控循环神经网络的海杂波幅度预测[J]. 电波科学学报, 2020, 35(2):257-263.
	MA L W, ZHANG J P, WU J J, et al. Prediction of sea clutter amplitude based on gated cyclic neural network[J]. Chinese Journal of Radio Science, 2020, 35(2):257-263. (in Chinese)
[7]	何耀民, 何华锋, 徐永壮, 等. 基于反向传播神经网络的海杂波参数估计[J]. 兵工学报, 2019, 40(12):2473-2481. doi: 10.3969/j.issn.1000-1093.2019.12.011
	HE Y M, HE H F, XU Y Z, et al. Sea clash parameter estimation based on back-propagation neural network[J]. Acta Armamentarii, 2019, 40(12):2473-2481. (in Chinese)
[8]	许桂桂. 基于深度学习的海面弱目标探测技术研究[D]. 武汉: 华中科技大学, 2019.
	XU G G. Research on weak target detection technology in sea surface based on deep learning[D]. Wuhan: Huazhong University of Science and Technology, 2019. (in Chinese)
[9]	施赛楠, 董泽远, 杨静, 等. 基于时频图自主学习的海面小目标检测[J]. 系统工程与电子技术, 2021, 43(1):33-41.
	SHI S N, DONG Z Y, YANG J, et al. Small target detection in sea surface based on time-frequency graph autonomous learning[J]. Systems Engineering and Electronics, 2021, 43(1):33-41. (in Chinese)
[10]	潘美艳, 孙俊, 杨予昊, 等. 基于Faster R-CNN网络的海面目标检测方法[J]. 现代雷达, 2021, 43(6):19-26.
	PAN M Y, SUN J, YANG Y H, et al. Sea surface target detection method based on faster R-CNN network[J]. Modern Radar, 2021, 43(6):19-26. (in Chinese)
[11]	CHEN Z Z, HE C, ZHAO C, et al. Using SVD-FRFT filtering to suppress first-order sea clutter in HFSWR[J]. IEEE Geoscience and Remote Sensing Letters, 2017, 14(7):1076-1080. doi: 10.1109/LGRS.2017.2697458 URL
[12]	王龙岗, 岳显昌, 吴雄斌, 等. 基于奇异值分解的空域海杂波抑制算法[J]. 电波科学学报, 2021, 36(4):645-652.
	WANG L G, YUE X C, WU X B, et al. Spatial sea clutter suppression algorithm based on singular value decomposition[J]. Chinese Journal of Radio Science, 2021, 36(4):645-652. (in Chinese)
[13]	孙国政, 卞雷祥. 雷达杂波自适应抑制技术[J]. 雷达与对抗, 2011, 31(2):11-13.
	SUN G Z, BIAN L X. Radar clutter adaptive suppression technology[J]. Radar and Countermeasure, 2011, 31(2):11-13. (in Chinese)
[14]	唐先慧, 李东, 粟嘉, 等. 基于AlexNet的自适应杂波智能抑制方法[J]. 信号处理, 2020, 36(12):2032-2042.
	TANG X H, LI D, SU J, et al. Adaptive clash suppression based on AlexNet[J]. Signal Processing, 2020, 36(12):2032-2042. (in Chinese)
[15]	BRUNA J, ZAREMBA W, SZLAM A, et al. Spectral networks and locally connected networks on graphs[J]. Computer Science, 2013, 29(11):1-14. doi: 10.1063/1.30584 URL
[16]	MIKOLOV T, CHEN K, CORRADO G, et al. Efficient estimation of word representations in vector space[C]// Proceedings of the 1st International Conference on Learning Representations.Scottsdale,AZ,US: 2013:1-9.
[17]	Grover Aditya and Leskovec Jure. node2vec: Scalable feature learning for networks[C]// Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.San Francisco,CA,US:ACM, 2016:855-864.
[18]	车转转. 海杂波中目标检测的时频方法研究[D]. 西安: 西安电子科技大学, 2014.
	CHE Z Z. Research on time-frequency method of target detection in sea cluttered[D]. Xi'an: Xidian University, 2014. (in Chinese)
[19]	PEROZZI B, AI-RFOU R, SKIENA S. Deepwalk:online learning of social representations[C]// Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.New York,NY,US:ACM, 2014:701-710.
[20]	TANG J, QU M, WANG M, et al. LINE:Large-scale information network embedding[C]// Proceedings of the 24th International Conference on World Wide Web.Florence,Italy:ACM, 2015:1067-1077.
[21]	杨晋吉, 胡波, 王欣明, 等. 一种知识图谱的排序学习个性化推荐算法[J]. 小型微型计算机系统, 2018, 39(11):2419-2423.
	YANG J J, HU B, WANG X M, et al. A sorting learning personalized recommendation algorithm for knowledge graph[J]. Journal of Small Microcomputer Systems, 2018, 39(11):2419-2423. (in Chinese)
[22]	雷雨. 用户属性信息推断方法的研究[D]. 哈尔滨: 哈尔滨工业大学, 2019.
	LEI Y. Research on user attribute information inference method[D]. Harbin:Harbin Institute of Technology, 2019. (in Chinese)
[23]	刘宇, 顾振杰, 韩红斌. 基于实测数据的海杂波建模方法[J]. 兵工自动化, 2019, 38(9):64-67.
	LIU Y, GU Z J, HAN H B. Sea clutter modeling method based on measured data[J]. Ordnance Industry Automation, 2019, 38(9):64-67. (in Chinese)
[24]	刘宁波, 丁昊, 黄勇, 等. X波段雷达对海探测试验与数据获取年度进展[J]. 雷达学报, 2021, 10(1):173-182.
	LIU N B, DING H, HUANG Y, et al. Annual progress of X-band radar sea-to-sea detection experiment and data acquisition[J]. Acta Radar Sinica, 2021, 10(1):173-182. (in Chinese)
[25]	关泽文, 陈建文, 鲍拯. 一种改进的基于峰值信噪比-高阶奇异值分解的天波超视距雷达自适应海杂波抑制算法[J]. 电子与信息学报, 2019, 41(7):1743-1750.
	GUAN Z W, CHEN J W, BAO Z. An improved adaptive sea clutter suppression algorithm for sky-wave over-the-horizon radar based on peak signal-to-noise ratio and high-order singular value decomposition[J]. Journal of Electronics & Information Technology, 2019, 41(7):1743-1750. (in Chinese)

参数	取值
目标回波信号中的目标个数K	{1,2,3}
目标回波信号的多普勒频率分量f_d/Hz	{50,75,100}
形状参数$\hat{a}$	0.7435
尺度参数$\hat{v}$	2.3281
超参数p和q	p∈{1,2,4},q∈{0.25,0.50}
节点特征向量的维度d	128
随机游走的次数r	10
分类参数n	{5,6,7}

参数	取值
目标回波信号中的目标个数K	{1,2,3}
目标回波信号的多普勒频率分量f_d/Hz	{50,75,100}
形状参数$\hat{a}$	0.7435
尺度参数$\hat{v}$	2.3281
超参数p和q	p∈{1,2,4},q∈{0.25,0.50}
节点特征向量的维度d	128
随机游走的次数r	10
分类参数n	{5,6,7}

相似度		多普勒频率/Hz
		50		75		100
		u₁	u₂	u₁	u₂	u₁	u₂
	{1,0.25}	0.62	0.29	0.63	0.21	0.94	0.07
	{2,0.25}	0.75	0.14	0.74	0.12	0.90	0.17
{p,q}	{4,0.25}	0.77	0.13	0.83	0.11	0.96	0.05
	{1,0.50}	0.65	0.22	0.69	0.20	0.73	0.18
	{2,0.50}	0.69	0.21	0.72	0.18	0.83	0.15
	{4,0.50}	0.75	0.25	0.80	0.17	0.87	0.14

相似度		多普勒频率/Hz
		50		75		100
		u₁	u₂	u₁	u₂	u₁	u₂
	{1,0.25}	0.62	0.29	0.63	0.21	0.94	0.07
	{2,0.25}	0.75	0.14	0.74	0.12	0.90	0.17
{p,q}	{4,0.25}	0.77	0.13	0.83	0.11	0.96	0.05
	{1,0.50}	0.65	0.22	0.69	0.20	0.73	0.18
	{2,0.50}	0.69	0.21	0.72	0.18	0.83	0.15
	{4,0.50}	0.75	0.25	0.80	0.17	0.87	0.14

相似度		多普勒频率/Hz
		50		75		100
		u₁	u₂	u₁	u₂	u₁	u₂
	{1,0.25}	0.60	0.30	0.62	0.23	0.91	0.08
	{2,0.25}	0.69	0.26	0.72	0.15	0.87	0.10
{p,q}	{4,0.25}	0.75	0.20	0.79	0.13	0.94	0.06
	{1,0.50}	0.64	0.25	0.62	0.29	0.72	0.20
	{2,0.50}	0.67	0.23	0.70	0.20	0.82	0.16
	{4,0.50}	0.71	0.22	0.79	0.19	0.75	0.21