基于复数卷积-残差网络的雷达杂波幅度统计模型分类

doi:10.12305/j.issn.1001-506X.2021.11.07

摘要/Abstract

摘要：

雷达杂波幅度统计模型分类是进行杂波背景下检测目标的重要步骤。雷达杂波原始数据通常是复数数据, 但现有杂波幅度统计模型分类研究都是在实数数据上完成的。复数数据同时包含幅度和相位信息, 更丰富的信息量有助于雷达杂波幅度统计模型分类。为此, 引入复数神经网络, 利用仿真杂波高分辨距离像(high resolution range profile, HRRP)复数数据, 对雷达杂波幅度统计模型分类问题进行研究, 完成了以下工作: 一是为构建复数最大池化层, 定义并改进了复数最大池化算法, 通过复数卷积神经网络(complex-valued convolutional neural networks, CV-CNN) 对杂波幅度统计模型的分类实验, 对比了两种复数最大池化算法和复数平均池化算法的分类效果, 实验结果表明复数最大池化算法的分类效果更好, 分类准确率为97.29%;二是为进一步提高分类准确率, 构建了复数卷积-残差网络(complex-valued convolution-ResNet, CV-CRN), 通过实验对比分析了CV-CRN的性能, 实验结果表明, CV-CRN的分类性能优于CV-CNN, 分类准确率达到98.84%, 并具有较好的鲁棒性。

关键词: 杂波分类, 高分辨距离像, 复数卷积-残差网络

Abstract:

The classification of clutter amplitude statistical model is an important step in detecting targets under clutter background. However, the raw clutter data from radar are usually complex numbers, and most of current research on classifying clutter amplitude statistical models focuses on real data. Complex data contain not only amplitude but phase information as well, and hence are beneficial to the classification problem we concern. In order to study the classification of radar clutter amplitude statistical model, we propose to use complex-valued neural network to process simulated complex clutter data of high resolution range profile (HRRP).The following work is completed: First, in order toconstructcomplex-valued pooling layer, the complex-valued maximum pooling algorithm is defined and improved based on python programming language, and the classification effect of two kinds of complex-valued maximum pooling algorithm and complex-valued average pooling algorithm is compared by using complex-valued convolutional neural networks (CV-CNN) to classify the statistical model of clutter amplitude. The experimental results show that the complex-valued maximum pooling algorithm is more effective and the classification accuracy reaches 97.29%. Second, in order to further improve the classification accuracy, we constructed the complex-valued convolution-ResNet (CV-CRN), then, compared and analyzed the performance of CV-CRN through the experiments of classification of clutter amplitude statistical model. Experimental results have revealed that, the classification performance of CV-CRN was better than that of CV-CNN, with classification accuracy rate being 98.84% and good robustness.

Key words: clutter classification, high resolution range profile (HRRP), complex-valued convolutional-resNet (CV-CRN)

中图分类号:

TN957.51

张良, 杨威, 李玮杰, 杨小琪, 刘永祥. 基于复数卷积-残差网络的雷达杂波幅度统计模型分类[J]. 系统工程与电子技术, 2021, 43(11): 3086-3097.

Liang ZHANG, Wei YANG, Weijie LI, Xiaoqi YANG, Yongxiang LIU. Classification of radar clutter amplitude statistical model based on complex-valued convolutional-ResNet[J]. Systems Engineering and Electronics, 2021, 43(11): 3086-3097.

图/表 14

图1

图2

图3

图4

图5

图6

表1

表2

图7

图8

图9

表3

表4

表5

参考文献 32

1	方学立, 马晓岩, 向家彬. 基于高阶统计量特征提取的雷达杂波分类识别[J]. 空军雷达学院学报, 2000, 14 (1): 9- 12.
	FANG X L , MA X Y , XIANG J B . Radar clutter recognition and higher-order statistics feature extraction[J]. Journal of the Air Force Radar Academy, 2000, 14 (1): 9- 12.
2	STEHWIEN W, HAYKIN S. aStatistical radar clutter classi-fier[C]//Proc. of the Radar Conference, 1989: 164-169.
3	HAYKIN S , KESLER S , CURRIE B . An experimental classification of radar clutter[J]. Proc.of the IEEE, 1979, 67 (2): 332- 333. doi: 10.1109/PROC.1979.11246
4	HAYKIN S , DENG C . Classification of radar clutter using neural networks[J]. IEEE Trans.on Neural Networks, 1991, 2 (6): 589- 600. doi: 10.1109/72.97936
5	靳珍璐, 潘泉, 梁彦, 等. 基于SVM的地海杂波多特征识别算法[C]//第31届中国控制会议论文集, 2012: 3903-3908.
	JIN Z L, PAN Q, LIANG Y, et al. SVM-based land/sea clutter identiifcation with multi-features[C]//Proc. of the 31st Chinese Control Conference, 2012: 3903-3908.
6	房永奎. 电离层杂波检测与污染程度评估研究[D]. 哈尔滨: 哈尔滨工业大学, 2010.
	FANG Y K. Study on ionospheric clutter detection and assessment of pollution level[D]. Harbin: Harbin Institute of Technology, 2010.
7	何炜琨, 郭双双, 王晓亮, 等. 基于Level-Ⅱ数据和模糊逻辑推理的气象雷达风电场杂波检测与识别方法[J]. 电子与信息学报, 2016, 38 (12): 3252- 3260.
	HE W K , GUO S S , WANG X L , et al. Weather radar wind farms clutters detection and identification method based on level-Ⅱ data and fuzzy logic inference[J]. Journal of Electronics and Information Technology, 2016, 38 (12): 3252- 3260.
8	张贤亮, 尚尚, 魏海峰. 基于LOWESS回归的一阶海杂波识别算法[J]. 电子器件, 2019, 42 (2): 458- 462. doi: 10.3969/j.issn.1005-9490.2019.02.037
	ZHANG X L , SHANG S , WEI H F . An algorithm for the first-order sea clutter's recognition based on LOWESS[J]. Chinese Journal of Electron Devices, 2019, 42 (2): 458- 462. doi: 10.3969/j.issn.1005-9490.2019.02.037
9	赵亮, 张友辉. 基于SCI算法的雷达地物杂波识别技术[J]. 民航学报, 2018, 2 (5): 91- 95.
	ZHAO L , ZHANG Y H . Research on radar ground object clutter recognition technology based on SCI algorithm[J]. Journal of Civil Aviation, 2018, 2 (5): 91- 95.
10	SHANG S, ZHANG N, LI Y. Extraction of ionospheric clutter in HFSWR[C]//Proc. of the 2nd International Conference on Industrial Mechatronics and Automation, 2010: 296-299.
11	LI Y, ZENG W J, ZHANG N, et al. Gabor feature based ionospheric clutter region extraction in Range-Doppler map[C]//Proc. of the IEEE Antennas and Propagation Society International Symposium, 2014: 269-270.
12	SILVEIRA R.B , HOLT a R . An automatic identification of clutter and anomalous propagation in polarization-diversity weather radar data using neural networks[J]. IEEE Trans.on geoscience and remote sensing, 2001, 39 (8): 1777- 1788.
13	ISMAIL N N, RASHID N E A, KHAN Z I, et al. Measurement, processing and modeling of atropical foliage clutter using forward scatter radar micro-sensor network with VHF and UHF bands[C]//Proc. of the International Conference on Radar, Antenna, Microwave, Electronics and Telecommunications, 2015: 76-81.
14	李旭涛, 王首勇, 金连文. 应用Alpha稳定分布对雷达杂波的辨识[J]. 电子与信息学报, 2008, 30 (9): 2042- 2045.
	LI X T , WANG S Y , JIN L W . Radar clutter recognition using alpha stable distribution[J]. Journal of Electronics & Information Technology, 2008, 30 (9): 2042- 2045.
15	NEINAVAIE M, DERAKHTIAN M, SHEIKHI A, et al. Clutter classification in heterogeneous environments[C]//Proc. of the 23rd Conference on Electrical and Compute Engineering, 2010.
16	贺梦珂. 基于机器学习的高频地波雷达复杂杂波识别技术研究[D]. 哈尔滨: 哈尔滨工业大学, 2017.
	HE M K. High frequency surface wave radar complex clutter recognition based on machine learning technology[D]. Harbin: Harbin Institute of Technology, 2017: 52-60.
17	Nitta T. On the critical points of the complex-valued neural network[C]//Proc. of the IEEE 9th International Conference on Neural Information Processing, 2002: 1099-1103.
18	ARJOVSKY M , SHAH A , BENGIO Y . Unitary evolution recurrent neural networks[J]. arXiv: 1511.06464v4[cs.LG], 2016, 1- 9.
19	WISDOM S , POWERS T , HERSHEY J , et al. Full-capacity unitary recurrent neural networks[J]. arXiv: 1611.00035[stat.ML], 2016, 1- 9.
20	Kim M S, Guest C C. Modification of backpropagation networks for complex-valued signal processing in frequency domain[C]// Proc. of the IJCNN International Joint Conference on Neural Networks, 1990: 27-31.
21	Clarke T L. Generalization of neural networks to the complex plane[C]//Proc. of the IJCNN International Joint Conference on Neural Networks, 1990: 435-440.
22	RONNY H, OLAF H. Classification of polarimetric SAR data by complex valued neural networks[C]//Proc. of the ISPRS Hannover Workshop, 2009.
23	ZHANG Z M, WANG H P, XU F, et al. Complex-valued convolutional neural network and its application in polarimetric SAR image classification[C]//Proc. of the IEEE Trans. on Geoscience and Remote Sensing, 2017: 7177-7188.
24	CHIHEB TI, OLEXA B, ZHANG Y, et al. Deep complex network[C]//arXiv: 1705.09792v4[cs. NE] 25 Feb 2018: 1-19.
25	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
26	SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2015.
27	高敬坤. 阵列三维成像及雷达增强成像技术研究[D]. 长沙: 国防科技大学, 2018.
	GAO J K. Research on array-based 3D imaging and enhaned radar imaging techniques[D]. Changsha: National University of Defense Technology, 2018.
28	GOODFELLOW I, BENGIO Y, COURVILLE A, 等. 深度学习[M]. 赵申剑, 黎彧君, 符天凡, 等译. 北京: 人民邮电出版社, 2017: 199-226.
	GOODFELLOW I, BENGIO Y, COURVILLE A, et al. Deep Learning[M]. ZHAO S J, LI Y J, FU T F, et al. Tran. Beijing: Posts and Telecommunications Press, 2017: 199-226.
29	ZHANG W , LI C H , PENG G L , et al. a deep convolutional neural network with new training methods for bearing fault diagnosis under noisy environment and different working load[J]. Mechanical Systems and Signal Processing, 2018, 100, 439- 453.
30	ZEILER M D, FERGUS R. Visualizing and understanding convolutional networks[C]//Proc. of the 13th European Conference on Computer, 2014: 818-833.
31	张宝宝. 雷达杂波的建模仿真技术研究[D]. 西安: 西安电子科技大学, 2007.
	ZHANG B B. Research on radar clutter's modeling and simulation[D]. Xi'an: Xidian University, 2007.
32	罗军辉, 罗勇江, 白义臣, 等. Matlab7.0在数字信号处理中的应用[M]. 北京: 机械工业出版社, 2005: 169- 184.
	LUO J H , LUO Y J , BAI Y C , et al. Application of Matlab7.0 in digital signal processing[M]. Beijing: China Machine Press, 2005: 169- 184.

分布类型	概率密度函数	参数说明
瑞利分布	$f(x;\sigma ) = \frac{x}{{{\sigma ^2}}}\exp ( - \frac{x}{{2{\sigma ^2}}}),x \ge 0 $	方差σ²
韦布尔分布	$f(x;a,b) = \frac{b}{{{a^b}}}{x^{b - 1}}\exp [ - {(\frac{x}{a})^b}],a > 0,b > 0,x \ge 0 $	形状参数b 尺度参数a
K分布	$f(x;a,b) = \frac{2}{{a\mathit{\boldsymbol{ \boldsymbol{\varGamma} }}(b)}}{(\frac{x}{{2a}})^b}{K_{b - 1}}(\frac{x}{a}),a > 0,b > 0,x \ge 0 $	形状参数b 尺度参数a
对数正态分布	$f(x;\mu ,\sigma ) = \frac{1}{{\sqrt {2\pi } \sigma x}}\exp [ - \frac{{{{(\ln x - \mu )}^2}}}{{2{\sigma ^2}}}],x \ge 0 $	均值μ 方差σ²

分布类别	参数	取值范围	备注
瑞利分布	σ	0.1~5	σ为标准差
韦布尔分布	b a	0.1~10 0.1~10	μ为均值 b为形状参数
对数正态分布	μ σ	0.1~5 0.1~5	a为尺度参数 K分布的尺度参数
K分布	b	0.1~10	通过非线性计算得到

网络模型	分类准确率
网络模型	平均	瑞利杂波	韦布尔杂波	K分布
CV-CNN-M1	97.29	96.60	93.95	98.60
CV-CNN-M2	96.80	96.05	92.60	98.65
CV-CNN-A	96.29	94.30	91.95	98.90
RV-CNN-M	95.38	92.90	89.65	99.15
RV-CNN-A	92.19	91.55	81.15	98.75

网络模型	分类准确率
网络模型	平均	瑞利杂波	韦布尔杂波	K分布	对数正态分布
CA-RM1	98.84	97.60	98.90	98.85	100
CA-RM2	98.26	97.45	96.75	98.85	100
CA-RA	96.46	95.75	92.00	98.15	99.95
CM1-RM1	98.31	96.40	97.20	99.65	100
CM1-RM2	96.83	93.05	93.50	98.65	100
CM1-RA	97.16	96.25	93.50	98.90	100
CM2-RM1	97.76	97.10	96.30	97.70	99.95
CM2-RM2	98.17	96.80	96.70	99.20	100
CM2-RA	96.45	93.95	94.05	97.80	100

杂噪比	网络
杂噪比	无噪	10 dB	5 dB	0 dB
CV-CRN	98.84	95.45	94.65	93.42
CV-CNN	97.29	95.53	94.50	93.19
RV-CNN	95.38	92.94	91.15	88.29