基于深度神经网络的水下分布源波达方向估计算法

梁奕念; 李杰; 陈芳炯; 季飞; 余华

doi:10.11993/j.issn.2096-3920.2024-0178

基于深度神经网络的水下分布源波达方向估计算法

doi: 10.11993/j.issn.2096-3920.2024-0178

华南理工大学电子与信息学院, 广东广州, 510641

基金项目: 国家自然科学基金(62271208; 62192712; 62341129); 广州市重点研发项目(2023B03J1308); 广东省基础与应用基础研究基金(2025A1515011040).

详细信息

作者简介:
梁奕念(1993-), 女, 在读博士, 主要研究方向为深度学习在水声阵列信号处理中的应用

中图分类号: TJ630, U663
计量
- 文章访问数: 77
- HTML全文浏览量: 31
- PDF下载量: 17
- 被引次数: 0
出版历程
- 收稿日期: 2024-12-24
- 修回日期: 2025-01-28
- 录用日期: 2025-02-08
- 网络出版日期: 2025-02-25

Direction of Arrival Estimation Algorithm for Underwater Distributed Sources Based on Deep Neural Network

School of Electronics and Information Engineering, South China University of Technology, Guangzhou 510641, China

摘要

摘要: 针对传统子空间类波达方向(DOA)估计算法在处理不同相干性分布源定位时需依赖先验相干性信息的局限, 文中提出一种基于深度神经网络(DNN)的水下分布源DOA估计方法。该方法利用部分相干分布源信号模型中时间和角度相干分量的可分性, 通过分段均值归一化方法简化模型, 并构建DNN模型，通过多组不同相干系数的样本数据训练, 实现了对不同相干性分布源DOA角度的鲁棒性估计。仿真实验结果表明, 该方法无需相干性先验知识即可有效估计不同相干系数下的分布源参数。将文中方法与4种传统子空间类方法和1种深度卷积神经网络算法进行对比, 结果表明：在相干分布源情况下，文中方法在不同信噪比和快拍数条件下的均方根误差(RMSE)结果比其他方法平均降低0.42°；在非相干分布源情况下，当信噪比大于0 dB且快拍数大于600时，文中方法的RMSE结果比其他方法平均降低0.04°；在全相干系数范围内，文中方法均表现出更优的估计性能，验证了其在复杂水下环境中的适用性。
- 深度神经网络 /
- 分布源 /
- 波达方向估计 /
- 目标定位
Abstract: In view of the limitation of traditional subspace-based direction of arrival(DOA) estimation algorithms that rely on prior coherence information for localizing distributed sources with varying coherences, a DOA estimation algorithm for underwater distributed sources based on deep neural network(DNN) was proposed in this paper. By leveraging the separability of temporal and angular coherence components in the partially coherent distributed source signal model and simplifying the model by segmented mean normalization, a DNN model was constructed and trained with multiple samples of different coherence coefficients, thus achieving robust DOA estimation for distributed sources with different coherence levels. Simulation results indicate that the proposed method can effectively estimate the distributed source parameters with different coherent coefficients without relying on coherence prior knowledge. The proposed method is compared with four traditional subspace-based methods and one deep convolutional neural network algorithm, and the results show that the root mean square error of the proposed method under the coherently distributed source case is 0.42° lower than that of other methods under different signal-to-noise ratios(SNRs) and snapshots; under the incoherently distributed source case, the RMSE of the proposed method is 0.04° lower than that of other methods with SNR greater than 0 dB and snapshots greater than 600. In the range of full coherence coefficients, the proposed methods all show better estimation performance, which proves their applicability in complex underwater environments.
- deep neural network /
- distributed source /
- direction of arrival estimation /
- target localization

HTML全文

图 1 基于DNN的水下分布源DOA估计算法模型结构图

Figure 1. Model structure of a DOA estimation algorithm for underwater distributed source based on DNN

下载: 全尺寸图片幻灯片

图 2 DNN模型训练迭代次数与损失值关系曲线

Figure 2. The relationship between the epochs and the loss value of DNN model

下载: 全尺寸图片幻灯片

图 3 不同SNR下ID源RMSE结果对比

Figure 3. The RMSE results of ID source versus different SNRs

下载: 全尺寸图片幻灯片

图 4 不同SNR下CD源RMSE结果对比

Figure 4. The RMSE results of CD source versus different SNRs

下载: 全尺寸图片幻灯片

图 5 不同快拍数下ID源RMSE结果对比

Figure 5. The RMSE results of ID source versus different snapshots

下载: 全尺寸图片幻灯片

图 6 不同快拍数下CD源RMSE结果对比图

Figure 6. The RMSE results of CD source versus different snapshots

下载: 全尺寸图片幻灯片

图 7 PCD源在不同相干系数下的RMSE结果对比图

Figure 7. The root mean square error results of PCD source versus different coherence coefficient $ \alpha $

下载: 全尺寸图片幻灯片

参考文献(20)

[1]	周烨, 温玮, 韩建辉, 等. 基于声固耦合的水下复杂目标声散射研究[J]. 水下无人系统学报, 2020, 28(1): 51-56. ZHOU Y, WEN W, HAN J H, et al. Research on acoustic scattering of underwater complicated target based on sound-solid coupling[J]. Journal of Unmanned Undersea Systems, 2020, 28(1): 51-56.
[2]	曹浩, 樊书宏, 周晶. 基于板块元法的鱼雷声散射特性分析[J]. 水下无人系统学报, 2022, 30(2): 184-189. doi: 10.11993/j.issn.2096-3920.2022.02.007 CAO H, FAN S H, ZHOU J, et al. Analysis of acoustic scattering characteristics of a torpedo based on planar elements method[J]. Journal of Unmanned Undersea Sys- tems, 2022, 30(2): 184-189. doi: 10.11993/j.issn.2096-3920.2022.02.007
[3]	杜金香, 许恒博, 祝鹏. 一种角度加权的最小二乘目标定位算法[J]. 水下无人系统学报, 2019, 27(5): 570-573. DU J X, XU H B, ZHU P. An angle weighted least squares algorithm for target localization[J]. Journal of Unmanned Undersea Systems, 2019, 27(5): 570-573.
[4]	周敏. 相干分布式目标的方位估计[J]. 水下无人系统学报, 2008, 16(1): 36-39. ZHOU M. Direction of arrival estimation for coherent distributed target sources[J]. Journal of Unmanned Undersea Systems, 2008, 16(1): 36-39.
[5]	VALAEE S, CHAMPAGNE B, KABAL P. Parametric localization of distributed sources[J]. IEEE Transactions on Signal Processing, 1995, 43(9), 2144-2153.
[6]	梁奕念, 李杰, 龙力榕, 等. 利用卷积神经网络的非相干分布式源定位方法[J]. 声学学报, 2024, 49(1): 38-48. doi: 10.12395/0371-0025.2022138 LIANG Y N, LI J, LONG L R, et al. Incoherently distributed sources localization using convolutional neural network[J]. ACTA Acustica, 2024, 49(1): 38-48. doi: 10.12395/0371-0025.2022138
[7]	RAICH R, GOLDBERG J, MESSER H. Bearing estimation for a distributed source: Modeling, inherent accuracy limitations and algorithms[J]. IEEE Transactions on Signal Processing, 2000, 48(2): 429-441.
[8]	MENG Y, STOICA P, WONG K M. Estimation of the directions of arrival of spatially dispersed signals in array processing[J]. IEEE Proceedings Radar, Sonar and Navigation, 1996, 143(1): 1-9.
[9]	梁浩, 崔琛, 代林, 等. 基于ESPRIT算法的L型阵列MIMO雷达降维DOA估计[J]. 电子与信息学报, 2015, 37(8): 1828-1835. LIANG H, CUI C, DAI L, et al. Reduced-dimensional DOA estimation based on ESPRIT algorithm in MIMO radar with L-shaped array[J]. Journal of Electronics & Information Technology, 2015, 37(8): 1828-1835.
[10]	王绪虎, 陈建峰, 叶永,等.基于ESPRIT算法的矢量水听器阵方位估计性能分析[C]//2012年水下复杂战场环境目标识别与对抗及仿真技术学术交流. 成都, 中国: 中国造船工程学会电子技术学术委员会, 2012: 30-37.
[11]	SHAHBAZPANAHI S, VALAEE S, BASTANI M H. Distributed source localization using ESPRIT algorithm[J]. IEEE Transactions on Signal Processing, 2001, 49 (10): 2169-2178.
[12]	CAO R, GAO F, ZHANG X. An angular parameter estimation method for incoherently distributed sources via generalized shift invariance[J]. IEEE Transaction Signal Processing, 2016, 64(17): 4493-4503. doi: 10.1109/TSP.2016.2557312
[13]	WANG Y, AHMED Z. Some new techniques of localization of spatially distributed sources[C]//2007 Conference Record of the Forty First Asilomar Conference on Signals, Systems and Computers. Pacific Grove, CA, USA: IEEE, 2007: 1807-1811.
[14]	ZOUBIR A, WANG Y, CHARGÉ P. Efficient subspace-based estimator for localization of multiple incoherently distributed sources[J]. IEEE Transaction Signal Processing, 2008, 56(2): 532-542.
[15]	PAPAGEORGIOU G K, SELLATHURAI M, ELDAR Y C. Deep networks for direction-of-arrival estimation in low SNR[J]. IEEE Transactions on Signal Processing, 2021, 69: 3714-3729.
[16]	FEINTUCH S, TABRIKIAN J, BILIK I, et al. Neural-network-based DOA estimation in the presence of non-gaussian interference[J]. IEEE Transactions on Aerospace and Electronic Systems, 2023, 60(1): 119-132.
[17]	TIAN Y, LIU S, LIU W, et al. Vehicle positioning with deep-learning-based direction-of-arrival estimation of incoherently distributed sources[J]. IEEE Internet of Things Journal, 2022, 9(20): 20083-20095. doi: 10.1109/JIOT.2022.3171820
[18]	WANG J, QUAN T, JIAO L, et al. DOA estimation of underwater acoustic array signal based on wavelet transform with double branch convolutional neural network[J]. IEEE Transactions on Vehicular Technology, 2023, 72(5): 5962-5972. doi: 10.1109/TVT.2022.3203034
[19]	GE S G, LI K, SITI N B M R. Deep learning approach in DOA estimation: A systematic literature review[J]. Mobile Information Systems, 2021(1): 1-14.
[20]	ZHANG C L, LUO J H, WEI X S, et al. In defense of fully connected layers in visual representation transfer[C]//18th Pacific-Rim Conference on Multimedia (PCM). Harbin, China: PCM, 2018: 807-817.