基于生成模型的三维波束形成图像压缩方法

doi:10.3969/j.issn.1004-132X.2025.07.015

摘要/Abstract

摘要： 针对通道压缩方法在高压缩率下导致DenseNet模型定位性能显著降低的问题，提出一种基于改进向量量化变分自编码器(VQ-VAE-2)模型的三维波束形成图像压缩 (3D-BFMC) 方法。先利用VQ-VAE-2模型的层级编码器将三维波束形成图压缩为向量化局部特征矩阵，再将该矩阵输入DenseNet模型实现三维定位。仿真结果表明，使用3D-BFMC方法压缩数据训练的DenseNet模型在定位精度、频率泛化性能、噪声鲁棒性上均优于通道压缩方法。单声源试验验证了3D-BFMC方法在真实环境中的有效性和可行性。

关键词: 波束形成, 数据压缩, 深度学习, 改进向量量化变分自编码器, 三维空间

Abstract: To address the significant degradation in localization performance of DenseNet model under high compression ratios caused by channel compression method, a 3D beamforming map compression(3D-BFMC) method was proposed based on VQ-VAE-2 model. The hierarchical encoder of the VQ-VAE-2 model was used to compress 3D beamforming maps into vectorized local feature matrices, and then the matrices were input into the DenseNet model to perform 3D localization. Simulation results show that DenseNet models trained on compressed data by the 3D-BFMC method have better localization accuracy, frequency generalization and noise robustness than those of outperform channel compression approaches. A single-source experiment confirms the effectiveness and feasibility of 3D-BFMC in real-world environments.

Key words: beamforming, data compression, deep learning, vector quantized-variational autoencoder-2(VQ-VAE-2), three-dimensional space

中图分类号:

TP391
TN912

赵昀杰, 贺岩松, 张志飞, 徐中明. 基于生成模型的三维波束形成图像压缩方法[J]. 中国机械工程, 2025, 36(07): 1520-1529.

ZHAO Yunjie, HE Yansong, ZHANG Zhifei, XU Zhongming. 3D Beamforming Map Compression Method Based on Generative Model[J]. China Mechanical Engineering, 2025, 36(07): 1520-1529.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.cmemo.org.cn/CN/10.3969/j.issn.1004-132X.2025.07.015

https://www.cmemo.org.cn/CN/Y2025/V36/I07/1520

参考文献

［1］ABDELREHEEM A, MOHAMED E M, ESMAIEL H. Location-based Millimeter Wave Multi-level Beamforming Using Compressive Sensing［J］. IEEE Communications Letters, 2018, 22(1):185-188.
［2］van den ENDE M P A, AMPUERO J P. Evaluating Seismic Beamforming Capabilities of Distributed Acoustic Sensing Arrays［J］. Solid Earth, 2021, 12(4):915-934.
［3］MALGOEZAR A, SNELLEN M, SIJTSMA P, et al. Improving Beamforming by Optimization of Acoustic Array Microphone Positions［C］∥6th Berlin Beamforming Conference. Berlin, 2016:BeBeC-2016-S5.
［4］CHIARIOTTI P, MARTARELLI M, CASTELLINI P. Acoustic Beamforming for Noise Source Localization—Reviews, Methodology and Applications［J］. Mechanical Systems and Signal Processing, 2019, 120:422-448.
［5］DBLER D, OCKER J, PUHLE C. On 3D-Beamforming in The Wind Tunnel［C］∥6th Berlin Beamforming Conference. Berlin, 2016:BeBeC-2016-S10.
［6］MERINO-MARTINEZ R, VON DEN HOFF B, MORATA D, et al. Three-Dimensional Acoustic Imaging Using Asynchronous Microphone Array Measurements［C］∥9th Berlin Beamforming Conference. Berlin, 2022:S8.
［7］BROOKS T, HUMPHREYS W. Three-dimensional Applications of DAMAS Methodology for Aeroacoustic Noise Source Definition［C］∥11th AIAA/CEAS Aeroacoustics Conference. Reston, 2005:2960.
［8］XENAKI A, JACOBSEN F, FERNANDEZ-GRANDE E. Improving the Resolution of Three-dimensional Acoustic Imaging with Planar Phased Arrays［J］. Journal of Sound and Vibration, 2012, 331(8):1939-1950.
［9］ZAN Ming, XU Zhongming, TANG Zhonghua, et al. Three-dimensional Deconvolution Beamforming Based on the Variable-scale Compressed Computing Grid［J］. Measurement, 2022, 205:112211.
［10］MA Wei, LIU Xun. Phased Microphone Array for Sound Source Localization with Deep Learning［J］. Aerospace Systems, 2019, 2(2):71-81.
［11］KUJAWSKI A, HEROLD G, SARRADJ E. A Deep Learning Method for Grid-free Localization and Quantification of Sound Sources［J］. The Journal of the Acoustical Society of America, 2019, 146(3):EL225.
［12］XU Pengwei, ARCONDOULIS E J G, LIU Yu. Acoustic Source Imaging Using Densely Connected Convolutional Networks［J］. Mechanical Systems and Signal Processing, 2021, 151:107370.
［13］LEE S Y, CHANG J, LEE S. Deep Learning-based Method for Multiple Sound Source Localization with High Resolution and Accuracy［J］. Mechanical Systems and Signal Processing, 2021, 161:107959.
［14］ZHAO Yunjie, HE Yansong, CHEN Hao, et al. Three-dimensional Grid-free Sound Source Localization Method Based on Deep Learning［J］. Applied Acoustics, 2025, 227:110261.
［15］陈昱, 项薇, 林文文, 等. 基于半监督学习双模型结构的注塑产品异常检测［J］. 中国机械工程, 2025, 36(3):576-583.
CHEN Yu, XIANG Wei, LIN Wenwen, et al. Abnormal Detection of Injection Molding Products Based on Semi-supervised Learning Dual-model Structure［J］. China Mechanical Engineering, 2025, 36(3):576-583.
［16］杨光友, 刘浪, 习晨博. 自适应辅助分类器生成式对抗网络样本生成模型及轴承故障诊断［J］. 中国机械工程, 2022, 33(13):1613-1621.
YANG Guangyou, LIU Lang, XI Chenbo. Bearing Fault Diagnosis Based on SA-ACGAN Data Generation Model［J］. China Mechanical Engineering, 2022, 33(13):1613-1621.
［17］KINGMA D P, WELLING M. Auto-Encoding Variational Bayes［J］. arXiv Preprint arXiv, 2013:1312.6114.
［18］SARRADJ E. Three-dimensional Acoustic Source Mapping with Different Beamforming Steering Vector Formulations［J］. Advances in Acoustics and Vibration, 2012, 2012(1):292695.
［19］HUANG Gao, LIU Zhuang, van der MAATEN L, et al. Densely Connected Convolutional Networks［C］∥2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, 2017:2261-2269.
［20］ALZUBAIDI L, ZHANGJinglan, HUMAIDI A J, et al. Review of Deep Learning:Concepts, CNN Architectures, Challenges, Applications, Future Directions［J］. Journal of Big Data, 2021, 8(1):53.
［21］PASZKE A, GROSS S, MASSA F, et al. PyTorch:An Imperative Style, High-Performance Deep Learning Library［J］. Advances in Neural Information Processing Systems, 2019, 32.
［21］PASZKE A, GROSS S, MASSA F, et al. PyTorch:an Imperative Style, High-performance Deep Learning Library［J］. arXiv, 2019:1912.01703v1.
［22］SARRADJ E, HEROLD G. A Python Framework for Microphone Array Data Processing［J］. Applied Acoustics, 2017, 116:50-58.
［23］KUJAWSKI A, PELLING A J R, JEKOSCH S, et al. A Framework for Generating Large-scale Microphone Array Data for Machine Learning［J］. Multimedia Tools and Applications, 2024, 83(11):31211-31231.
［24］VAN DEN OORD A, VINYALS O. Neural Discrete Representation Learning［J］. arXiv, 2017:1711.00937.
［25］RAZAVI A, VAN DEN OORD A, VINYALS O. Generating Diverse High-Fidelity Images with VQ-VAE-2［J］. arXiv, 2019:1906.00446.
［26］KINGMA D P, BA J. Adam:a Method for Stochastic Optimization［J］. arXiv, 2014:1412.6980.
［27］LOSHCHILOV I, HUTTER F. SGDR:Stochastic Gradient Descent with Warm Restarts［J］. arXiv, 2016:1608.03983.

[1]	曾浩, 曹华军, 董俭雄. 基于ISABO-IBiLSTM模型的刀具磨损预测方法[J]. 中国机械工程, 2024, 35(11): 1995-2006.
[2]	李悦1, 2, 谢恒1, 周公博1, 2, 周坪1, 2, 李猛钢1, 2. 基于半监督贝叶斯Transformer的刀具磨损软测量及不确定性分析方法[J]. 中国机械工程, 2024, 35(11): 2015-2025.
[3]	聂昕, 谭天, 申丹凤. 基于深度学习的汽车梁类件冲压回弹研究[J]. 中国机械工程, 2023, 34(07): 838-846.
[4]	唐东林, 杨洲, 程衡, 刘铭璇, 周立, 丁超. 浅层卷积神经网络融合Transformer的金属缺陷图像识别方法[J]. 中国机械工程, 2022, 33(19): 2298-2305，2316.
[5]	于浩, 黄华贵, 郑加丽, 赵铁琳, 周新亮. 无缝钢管斜轧穿孔顶头表面缺陷非接触在线检测方法[J]. 中国机械工程, 2022, 33(14): 1717-1724.
[6]	杨光友, 刘浪, 习晨博. 自适应辅助分类器生成式对抗网络样本生成模型及轴承故障诊断[J]. 中国机械工程, 2022, 33(13): 1613-1621.
[7]	艾青波, 张洁, 程辉, 吕佑龙, 左丽玲, 胡蓝. 基于PointCPP-LSF方法的航天筒段薄壁件滚弯成形质量分析[J]. 中国机械工程, 2022, 33(08): 977-985.
[8]	张胜文, 周曦, 李滨城, , 程德俊, , 陈文笛. 基于图像深度学习的零件加工特征信息提取方法[J]. 中国机械工程, 2022, 33(03): 348-355.
[9]	崔海华, 徐振龙, 杨亚鹏, 孟亚云, 王宝俊. 一种基于深度学习图像超分的环形靶标稳定检测方法[J]. 中国机械工程, 2021, 32(23): 2861-2867.
[10]	闫杰琼, 周来水, 胡少乾, 文思扬. 航空发动机外形点云的保特征去噪方法[J]. 中国机械工程, 2021, 32(23): 2850-2860，2889.
[11]	童靳于, 罗金, 郑近德. 基于增强深度自编码网络的滚动轴承故障诊断方法[J]. 中国机械工程, 2021, 32(21): 2617-2624.
[12]	尹晨, 周世超, 何建樑, 孙宇昕, 王禹林. 基于多源同步信号与深度学习的刀具磨损在线识别方法[J]. 中国机械工程, 2021, 32(20): 2482-2491.
[13]	余浩帅, 汤宝平, 张楷, 谭骞, 魏静. 小样本下混合自注意力原型网络的风电齿轮箱故障诊断方法[J]. 中国机械工程, 2021, 32(20): 2475-2481.
[14]	赵志宏, 李乐豪, 杨绍普, 李晴. 一种频域特征提取自编码器及其在故障诊断中的应用研究[J]. 中国机械工程, 2021, 32(20): 2468-2474.
[15]	戴稳1;张超勇1;孟磊磊1;薛燕社1;肖鹏飞1;尹勇2. 采用深度学习的铣刀磨损状态预测模型[J]. 中国机械工程, 2020, 31(17): 2071-2078.