-
摘要:
高频地波雷达因其卓越的海面目标探测能力，被世界各国应用于海上工程领域，而提升其目标探测能力的关键要素之一在于回波信号中电离层杂波的抑制，针对这一现象，提出一种基于瓶颈膨胀卷积模块改进时序卷积(ITCN)-Elman神经网络结合混合注意力机制的电离层杂波预测抑制模型(Mixatt-ITCN-Elman)。对电离层杂波时间序列进行相空间重构和乱序归一化，利用ITCN提取高维相空间内的空间特征，依据自注意力机制突出其中关键的空间特征，将空间特征与原时间序列组合输入Elman神经网络，结合注意力机制突显序列的空时特征，通过空时特征与Elman神经网络输出序列组合输出，得到最终预测结果。所提模型与Elman、TCN、Att-CNN-Elman和TCN-Elman模型相对比，具有较好的预测性能和稳定性，对于电离层杂波的抑制具有较高应用价值。
Abstract:High-frequency surface wave radar is used in the field of offshore engineering worldwide because of its excellent sea surface target detection capability. One of the key elements to improve its target detection capability lies in the suppression of ionospheric clutter in the echo signal. In response, this paper proposed an ionospheric clutter prediction and suppression model (Mixatt-ITCN-Elman) based on bottleneck expansion convolution module improved temporal convolution (ITCN)-Elman neural network combined with hybrid attention mechanism. First, the ionospheric clutter time series was reconstructed in phase space and subjected to disordered normalization. The spatial features within the high-dimensional phase space were extracted by using ITCN, and the key spatial features were highlighted by combining the self-attention mechanism. Then, the spatial features were combined with the original time series and input into the Elman neural network. The spatial-temporal features of the sequences were highlighted by combining the attention mechanism. Finally, the spatial-temporal features combined with the Elman neural network output sequence were output to obtain the final prediction result. In comparison with Elman, TCN, Att-CNN-Elman, and TCN-Elman models, the proposed model has better prediction performance and stability, having high application value for the suppression of ionospheric clutter.
-
图 10 实测数据电离层杂波抑制前后群距离-多普勒谱对比
Figure 10. Comparison of group distance-Doppler spectra before and after ionospheric clutter suppression of measured data
表 1 模型训练误差指标对比
Table 1. Comparison of model training error metrics
模型 R M S/% $ \rho $ Elman 0.0530 0.0426 7.203 8 0.968 2 TCN 0.0305 0.0203 3.895 8 0.993 2 Att-CNN-Elman 0.0208 0.0157 2.543 2 0.997 2 TCN-Elman 0.0201 0.0138 2.155 1 0.997 4 Mixatt-ITCN-Elman 0.0100 0.0076 1.145 5 0.999 2 
下载: 导出CSV
表 2 模型训练时间和预测精度对比
Table 2. Comparison of model training time and prediction accuracy
模型 训练时间/s 预测精度/% Elman 6.28 90.56 TCN 14.74 96.88 Att-CNN-Elman 15.96 98.55 TCN-Elman 20.42 98.64 Mixatt-ITCN-Elman 18.70 99.66
下载: 导出CSV
表 3 模型不同距离单元信杂比对比
Table 3. Comparison of signal-to-clutter ratio for different distance units of the model
模型 信杂比/dB 66距离单元 86距离单元 106距离单元 原始 −2.96 1.14 9.27 Elman 5.82 10.22 17.65 TCN 11.30 19.21 25.19 Att-CNN-Elman 12.53 17.15 24.72 TCN-Elman 12.58 19.70 24.83 Mixatt-ITCN-Elman 15.55 25.81 30.51
下载: 导出CSV
-
[1] COOK T M, TERRILL E J, GARCIA-MORENO C, et al. Observations of ionospheric clutter at near equatorial high frequency radar stations[J]. Remote Sensing, 2023, 15(3): 603. doi: 10.3390/rs15030603 [2] YANG X G, WANG M J, HUANG W M, et al. Experimental observation and analysis of ionosphere echoes in the mid-latitude region of China using high-frequency surface wave radar and ionosonde[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13: 4599-4606. doi: 10.1109/JSTARS.2020.3014666 [3] CHEN Z Z, XIE F, ZHAO C, et al. An orthogonal projection algorithm to suppress interference in high-frequency surface wave radar[J]. Remote Sensing, 2018, 10(3): 403. doi: 10.3390/rs10030403 [4] WANG Z Q, LI Y J, SHI J N, et al. Spread sea clutter suppression in HF hybrid sky-surface wave radars based on general parameterized time-frequency analysis[J]. International Journal of Antennas and Propagation, 2020, 2020: 1-12. [5] JANGAL F, SAILLANT S, HELIER M. Ionospheric clutter mitigation using one-dimensional or two-dimensional wavelet processing[J]. IET Radar, Sonar & Navigation, 2009, 3(2): 112-121. [6] DING M K, WEI Y S, YU L, et al. Adaptive suppression of main-lobe spread Doppler clutter with high directivity for HFSSWR using oblique projection[J]. Electronics Letters, 2019, 55(23): 1245-1247. doi: 10.1049/el.2019.1583 [7] LI J M, YANG Q, ZHANG X. Space-time adaptive processing algorithm based on hyper beamforming for ionospheric clutter suppression in small-array high-frequency surface wave radar[J]. IET Radar, Sonar & Navigation, 2023, 17(4): 545-555. [8] LI Y J, WANG Z Q, XU L, et al. Spread sea clutter suppression via prior knowledge-based space time adaptive processing in high frequency hybrid sky-surface wave radar[J]. IET Radar, Sonar & Navigation, 2023, 17(5): 830-844. [9] YANG Y L, MAO X P, HOU Y G, et al. A two-step method for ionospheric clutter mitigation for HFSWR with two-dimensional dual-polarized received array[J]. IEEE Access, 2020, 8: 105903-105913. doi: 10.1109/ACCESS.2020.2999463 [10] GENG H, LIANG Y, YANG F, et al. Joint estimation of target state and ionospheric height bias in over-the-horizon radar target tracking[J]. IET Radar, Sonar & Navigation, 2016, 10(7): 1153-1167. [11] CHEN S Y, HUANG W M, GILL E W. A vertical reflection ionospheric clutter model for HF radar used in coastal remote sensing[J]. IEEE Antennas and Wireless Propagation Letters, 2015, 14: 1689-1693. doi: 10.1109/LAWP.2015.2419174 [12] CHEN S Y, GILL E W, HUANG W M. A high-frequency surface wave radar ionospheric clutter model for mixed-path propagation with the second-order sea scattering[J]. IEEE Transactions on Antennas and Propagation, 2016, 64(12): 5373-5381. doi: 10.1109/TAP.2016.2618538 [13] 位寅生, 周建宇, 许荣庆. 基于杂波聚类与贪婪策略的电离层杂波智能处理方法[J]. 雷达学报, 2020, 9(4): 589-607. doi: 10.12000/JR20086WEI Y S, ZHOU J Y, XU R Q. Intelligent suppression method for ionospheric clutter based on clustering and greedy strategy[J]. Journal of Radars, 2020, 9(4): 589-607(in Chinese). doi: 10.12000/JR20086 [14] GENG H, LIANG Y, CHENG Y H. Target state and Markovian jump ionospheric height bias estimation for OTHR tracking systems[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2020, 50(7): 2599-2611. doi: 10.1109/TSMC.2018.2822819 [15] LYU Z, YU C, LIU A . Chaotic dynamics of ionospheric clutter from high frequency surface wave radar[C]//IET International Radar Conference. London: IET, 2020: 1640-1646. [16] CHENG W, WANG Y, PENG Z, et al. High-efficiency chaotic time series prediction based on time convolution neural network[J]. Chaos, Solitons & Fractals, 2021, 152: 111304. [17] HUANG W J, LI Y T, HUANG Y. Prediction of chaotic time series using hybrid neural network and attention mechanism[J]. Acta Physica Sinica, 2021, 70(1): 010501. doi: 10.7498/aps.70.20200899 [18] WEI L, CAO H, MA Z H. Parameter prediction of marine seawater cooling system based on chaos-Elman combined model[J]. IEEE Access, 2022, 10: 77272-77283. doi: 10.1109/ACCESS.2022.3183987 [19] WOLF A, SWIFT J B, SWINNEY H L, et al. Determining Lyapunov exponents from a time series[J]. Physica D: Nonlinear Phenomena, 1985, 16(3): 285-317. doi: 10.1016/0167-2789(85)90011-9 [20] KIM H S, EYKHOLT R, SALAS J D. Nonlinear dynamics, delay times, and embedding windows[J]. Physica D: Nonlinear Phenomena, 1999, 127(1-2): 48-60. doi: 10.1016/S0167-2789(98)00240-1 [21] HAYKIN S. Radar clutter attractor: implications for physics, signal processing and control[J]. IEE Proceedings-Radar, Sonar and Navigation, 1999, 146(4): 177-188. doi: 10.1049/ip-rsn:19990403 [22] 陈永, 陈锦, 陶美风. 多尺度特征和注意力融合的生成对抗壁画修复[J]. 北京麻豆精品秘 国产传媒学报, 2021, 49(2): 254-264.CHEN Y, CHEN J, TAO M F. Mural inpainting with generative adversarial networks based on multi-scale feature and attention fusion[J]. Journal of Beijing University of Aeronautics and Astronautics, 2023, 49(2): 254-264(in Chinese). [23] ZHU W Q, WANG Z Y, WANG X C, et al. A Dual self-attention mechanism for vehicle re-Identification[J]. Pattern Recognition, 2023, 137: 109258. doi: 10.1016/j.patcog.2022.109258 [24] KIM D, KIM Y. Understanding effects of architecture design to invariance and complexity in deep neural networks[J]. IEEE Access, 2021, 9: 9670-9681. doi: 10.1109/ACCESS.2021.3049841 [25] XING W Y, BAI Y L, DING L, et al. Application of a hybrid model based on GA–ELMAN neural networks and VMD double processing in water level prediction[J]. Journal of Hydroinformatics, 2022, 24(4): 818-837. doi: 10.2166/hydro.2022.016 [26] SUN Y Z, ZHANG J H, YU Z J, et al. WOA (whale optimization algorithm) optimizes Elman neural network model to predict porosity value in well logging curve[J]. Energies, 2022, 15(12): 4456. doi: 10.3390/en15124456 -
下载:
点击查看大图
计量
- 文章访问数: 164
- HTML全文浏览量: 75
- PDF下载量: 3
- 被引次数: 0
