基于深度学习的海杂波谱参数预测与影响因素分析

张玉石 李笑宇 张金鹏 夏晓云

张玉石, 李笑宇, 张金鹏, 等. 基于深度学习的海杂波谱参数预测与影响因素分析[J]. 雷达学报, 待出版. doi: 10.12000/JR22133
引用本文: 张玉石, 李笑宇, 张金鹏, 等. 基于深度学习的海杂波谱参数预测与影响因素分析[J]. 雷达学报, 待出版. doi: 10.12000/JR22133
ZHANG Yushi, LI Xiaoyu, ZHANG Jinpeng, et al. Sea clutter spectral parameters prediction and influence factor analysis based on deep learning[J]. Journal of Radars, in press. doi: 10.12000/JR22133
Citation: ZHANG Yushi, LI Xiaoyu, ZHANG Jinpeng, et al. Sea clutter spectral parameters prediction and influence factor analysis based on deep learning[J]. Journal of Radars, in press. doi: 10.12000/JR22133

基于深度学习的海杂波谱参数预测与影响因素分析

doi: 10.12000/JR22133
基金项目: 国家自然科学基金(U2006207)
详细信息
    作者简介:

    张玉石,博士,研究员,主要研究方向为地海杂波测试方法与系统、杂波数据分析与建模、智能认知与应用

    李笑宇,硕士生,主要研究方向为海杂波特性智能化处理与认知

    张金鹏,博士,高级工程师,主要研究方向为海杂波特性智能认知、电磁散射理论

    夏晓云,博士生,工程师,主要研究方向为海杂波特性智能认知、雷达海上目标检测

    通讯作者:

    张金鹏 zhangjp@crirp.ac.cn

  • 责任主编:刘宁波 Corresponding Editor: LIU Ningbo
  • 中图分类号: TN957

Sea Clutter Spectral Parameters Prediction and Influence Factor Analysis Based on Deep Learning

Funds: The National Natural Science Foundation of China (U2006207)
More Information
  • 摘要: 该文基于不同雷达参数和海洋环境参数条件下的岸基雷达海杂波实测数据,利用深度神经网络(DNN)建模技术,建立了从多个测量条件参数出发的海杂波多普勒谱参数预测模型,实现了独立于杂波数据、基于环境特征的海杂波谱特征认知,谱频移和展宽的预测精度达90%以上。基于该预测模型,该文提出了一种基于参数循环递减认知的多普勒谱影响因素分析方法,分析了不同测量参数对海杂波多普勒谱预测的影响,得到了谱参数随主要影响因素的变化规律,结果对基于多普勒特征的海面目标检测应用具有重要意义。

     

  • 图  1  Ku波段海杂波测量雷达实况

    Figure  1.  Location and reality of Ku band sea clutter measurement radar

    图  2  海杂波多普勒谱参数的提取流程

    Figure  2.  Extraction process of Doppler spectral parameters of sea clutter

    图  3  试验数据中风浪流方向的关系

    Figure  3.  Relationship between wind, wave and current direction in test data

    图  4  海杂波谱参数预测模型架构

    Figure  4.  Prediction model framework of sea clutter spectral parameters

    图  5  DNN-5网络结构图

    Figure  5.  DNN-5 network structure diagram

    图  6  基于DNN-5模型的谱参数预测散点图

    Figure  6.  Scatter plot of spectral parameter prediction based on DNN-5 model

    图  7  基于DNN-5模型的频移谱宽预测误差分布图

    Figure  7.  Error distribution of Doppler parameter prediction based on DNN-5 model

    图  8  基于循环递减认知的谱参数影响分析流程

    Figure  8.  Influence analysis process of spectral parameters based on cyclic decreasing cognition

    图  9  多普勒参数影响因子统计图

    Figure  9.  Statistical chart of influence factors of Doppler parameters

    图  10  主影响因素变化趋势拟合图

    Figure  10.  Fitting diagram of change trend of main influencing factors

    表  1  Ku波段海杂波数据集信息

    Table  1.   Ku band sea clutter dataset information

    海情等级HH极化(组)VV极化(组)总数(组)
    1186751616534840
    2283562372652082
    369421149918441
    合计5397351390105363
    下载: 导出CSV

    表  2  海杂波谱参数预测训练数据集组成形式

    Table  2.   Composition of training data set for prediction of sea clutter spectral parameters

    参数符号
    输入擦地角$ \theta $
    极化Pol
    带宽B
    有效浪高$ {H_{{\text{ave}}}} $
    浪向$ \alpha $
    浪周期T
    流速$ {v_{\text{f}}} $
    流向$ \beta $
    风速${v_{\rm{s}}}$
    风向$ \gamma $
    输出频移$ {f_{\text{d}}} $
    展宽$ {B_{\text{w}}} $
    下载: 导出CSV

    表  3  不同网络的隐层参数设置

    Table  3.   Hidden layer parameter settings of different networks

    模型名称隐藏层数激活函数损失函数优化函数
    DNN-55Leaky ReLU改进
    Huber Loss
    Adam函数
    BPNN2ReLU
    RBFNN2RBF
    RNN4ReLU
    SVR无须训练
    KNN
    下载: 导出CSV

    表  4  各个模型的谱参数预测结果

    Table  4.   Prediction results of spectral parameters of each model

    模型MAERMSE$ \eta $PR
    $ {f_{\text{d}}} $$ {B_{\text{w}}} $$ {f_{\text{d}}} $$ {B_{\text{w}}} $$ {f_{\text{d}}} $$ {B_{\text{w}}} $$ {f_{\text{d}}} $$ {B_{\text{w}}} $
    SVR17.465.1022.376.780.8150.8920.7490.780
    KNN9.875.0612.636.690.8890.9110.9260.792
    DNN-58.504.5811.576.160.9130.9530.9410.827
    BPNN11.974.8316.056.530.8770.9410.8770.802
    RBFNN9.154.6012.346.200.9050.9510.9300.819
    RNN10.264.6713.386.170.8870.9520.9220.826
    下载: 导出CSV

    表  5  基于循环递减认知方法的DNN-5评价指标

    Table  5.   Evaluation index of DNN-5 based on the cyclic decreasing cognition method

    去掉参数多普勒频移多普勒展宽
    MAERMSEPRη$ \varphi $MAERMSEPRη$ \varphi $
    擦地角9.6213.480.9160.9021.574.876.550.8030.9490.418
    极化方式9.0212.200.9370.9071.294.776.390.8280.9500.387
    带宽8.9012.030.9420.9091.254.936.620.8240.9470.418
    浪高9.1712.230.9420.9061.314.946.600.8230.9470.418
    浪向9.1512.250.9430.9061.314.916.590.8180.9470.417
    浪周期8.9011.830.9420.9091.234.906.510.8270.9480.407
    流速9.3012.560.9370.9041.384.906.570.8240.9480.412
    流向8.7611.770.9440.9101.204.906.570.8260.9480.411
    风速9.6313.110.9310.9001.514.866.620.8170.9480.415
    风向9.5412.750.9320.9021.455.086.790.8190.9450.446
    下载: 导出CSV
  • [1] SHI Yanling, GUO Yaxing, YAO Tingting, et al. Sea-surface small floating target recurrence plots FAC classification based on CNN[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5115713. doi: 10.1109/TGRS.2022.3192986
    [2] QU Qizhe, WANG Yongliang, LIU Weijian, et al. A false alarm controllable detection method based on CNN for sea-surface small targets[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 4025705. doi: 10.1109/LGRS.2022.3190865
    [3] WARD K D, TOUGH R J A, and WATTS S. Sea Clutter: Scattering, The K Distribution and Radar Performance[M]. 2nd ed. London: The Institution of Engineering and Technology, 2013: 129–142.
    [4] 李清亮, 尹志盈, 朱秀芹, 等. 雷达地海杂波测量与建模[M]. 北京: 国防工业出版社, 2017: 312–408.

    LI Qingliang, YIN Zhiying, ZHU Xiuqin, et al. Measurement and Modeling of Radar Clutter from Land and Sea[M]. Beijing: National Defense Industry Press, 2017: 312–408.
    [5] 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. doi: 10.12000/JR16069

    DING Hao, DONG Yunlong, LIU Ningbo, et al. Overview and prospects of research on sea clutter property cognition[J]. Journal of Radars, 2016, 5(5): 499–516. doi: 10.12000/JR16069
    [6] CROMBIE D D. Doppler spectrum of sea echo at 13.56 Mc. /s.[J]. Nature, 1955, 175(4459): 681–682. doi: 10.1038/175681a0
    [7] LEE P H Y, BARTER J D, BEACH K L, et al. Power spectral lineshapes of microwave radiation backscattered from sea surfaces at small grazing angles[J]. IEE Proceedings-Radar, Sonar and Navigation, 1995, 142(5): 252–258. doi: 10.1049/ip-rsn:19952084
    [8] WALKER D. Experimentally motivated model for low grazing angle radar Doppler spectra of the sea surface[J]. IEE Proceedings-Radar, Sonar and Navigation, 2000, 147(3): 114–120. doi: 10.1049/ip-rsn:20000386
    [9] ROSENBERG L and STACY N J. Analysis of medium grazing angle X-band sea-clutter Doppler spectra[C]. 2008 IEEE Radar Conference, Rome, Italy, 2008: 1–6.
    [10] ROSENBERG L. Parametric modeling of sea clutter Doppler spectra[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5105409. doi: 10.1109/TGRS.2021.3107950
    [11] ZHANG Jinpeng, ZHANG Yushi, XU Xinyu, et al. Estimation of sea clutter inherent Doppler spectrum from shipborne S-band radar sea echo[J]. Chinese Physics B, 2020, 29(6): 068402. doi: 10.1088/1674-1056/ab888a
    [12] WEN Liwu, DING Jinshan, ZHONG Chao, et al. Modeling of correlated complex sea clutter using unsupervised phase retrieval[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(1): 228–239. doi: 10.1109/TGRS.2020.2995892
    [13] MCDONALD M K and CERUTTI-MAORI D. Clairvoyant radar sea clutter covariance matrix modelling[J]. IET Radar, Sonar & Navigation, 2017, 11(1): 154–160. doi: 10.1049/iet-rsn.2016.0103
    [14] SHEN Yan and LI Guoqiang. The chaotic neural network is used to predict the sea clutter signal[C]. 2009 International Conference on Artificial Intelligence and Computational Intelligence, Shanghai, China, 2009: 25–30.
    [15] FERNÁNDEZ J R M and DE LA CONCEPCIÓN BACALLAO VIDAL J. Improved shape parameter estimation in K clutter with neural networks and deep learning[J]. International Journal of Interactive Multimedia and Artificial Intelligence, 2016, 3(7): 96–103. doi: 10.9781/ijimai.2016.3714
    [16] MA Liwen, WU Jiaji, ZHANG Jinpeng, et al. Sea clutter amplitude prediction using a Long short-term memory neural network[J]. Remote Sensing, 2019, 11(23): 2826. doi: 10.3390/rs11232826
    [17] SHUI Penglang, SHI Xiaofan, LI Xin, et al. GRNN-based predictors of UHF-band sea clutter reflectivity at low grazing angle[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 1502205. doi: 10.1109/LGRS.2021.3076842
    [18] 张晓峰, 王莉, 殷国东, 等. Ku波段地海杂波极化特性实验与分析[J]. 电波科学学报, 2019, 34(6): 676–686. doi: 10.13443/j.cjors.2019043003

    ZHANG Xiaofeng, WANG Li, YIN Guodong, et al. The experiments and analysis of polarization characteristics of the ground and sea clutters at Ku band[J]. Chinese Journal of Radio Science, 2019, 34(6): 676–686. doi: 10.13443/j.cjors.2019043003
    [19] MADSEN K, NIELSEN H B, and TINGLEFF O. Methods for Non-Linear Least Squares Problems[M]. 2nd ed. Lyngby: Informatics and Mathematical Modelling, Technical University of Denmark, 2004: 24–29.
  • 加载中
图(10) / 表(5)
计量
  • 文章访问数:  236
  • HTML全文浏览量:  74
  • PDF下载量:  67
  • 被引次数: 0
出版历程
  • 收稿日期:  2022-07-01
  • 修回日期:  2022-09-05
  • 网络出版日期:  2022-09-27

目录

    /

    返回文章
    返回