基于稀疏贝叶斯学习的压缩球波束形成声源识别方法

doi:10.3901/JME.2023.02.001

机械工程学报 ›› 2023, Vol. 59 ›› Issue (2): 1-13.doi: 10.3901/JME.2023.02.001

扫码分享

基于稀疏贝叶斯学习的压缩球波束形成声源识别方法

褚志刚^1,2, 赵洋¹, 杨亮^1,2, 张晋源³, 杨洋^1,3

1. 重庆大学机械与运载工程学院重庆 400044;
2. 汽车振动噪声和安全技术国家重点实验室重庆 401133;
3. 重庆工业职业技术学院车辆工程学院重庆 401120

收稿日期:2022-01-12 修回日期:2022-10-20 发布日期:2023-03-30
通讯作者: 杨亮(通信作者),男,1981年出生,博士研究生。主要研究方向为汽车振动噪声分析与控制技术。E-mail:yangliang2@changan.com.cn
作者简介:褚志刚,男,1978年出生,博士,教授,博士研究生导师。主要研究方向为传声器阵列信号处理、车辆振动噪声分析与控制。E-mail:zgchu@cqu.edu.cn;赵洋,男,1997年出生,博士研究生。主要研究方向为阵列声源识别技术、阵列信号处理。E-mail:zy970330@cqu.edu.cn;张晋源,男,1981年出生,教授。主要研究方向为汽车振动噪声分析与故障检测。E-mail:41540297125@qq.com;杨洋,女,1988年出生,副教授。主要研究方向为噪声源识别技术理论及其应用、阵列信号处理。E-mail:yangyang911127@cqu.edu.com
基金资助:
国家自然科学基金（11774040）和汽车噪声振动和安全技术国家重点实验室2022年度开放基金（NVHSKL-202202）资助项目。

Compressive Spherical Beamforming Based on Sparse Bayesian Learning for Sound Source Identification

CHU Zhigang^1,2, ZHAO Yang¹, YANG Liang^1,2, ZHANG Jinyuan³, YANG Yang^1,3

1. College of Mechanical and Vehicle Engineering, Chongqing University, Chongqing 400044;
2. State Key Laboratory of Vehicle NVH and Safety Technology, Chongqing 401133;
3. Faculty of Vehicle Engineering, Chongqing Industry Polytechnic College, Chongqing 401120

Received:2022-01-12 Revised:2022-10-20 Published:2023-03-30

摘要/Abstract

摘要： 基于实心球传声器阵列的压缩球波束形成具有声学成像全景、适宜中远距离测量而易于布置等优势，在汽车、飞机等噪声源识别领域具有广阔应用前景。新近提出的基于稀疏贝叶斯学习（Sparse Bayesian learning，SBL）的压缩球波束形成能够获得良好的低频声源识别性能，但由于其采用了第二类最大似然估计（Maximum type-Ⅱ likelihood estimation，MLE-Ⅱ）进而需要估计声源稀疏度，且抗噪声干扰能力和计算效率也有待提升，推广应用受限。为此，首先将压缩球波束形成数学模型求解问题转化为SBL框架下的源强分布最大后验（Maximum a posterior，MAP）估计问题，并采用期望最大化优化算法（Expectation maximization，EM）加以求解，提出无需稀疏度估计的MAP-EM压缩球波束形成方法；在此基础上，将多快拍复声压矩阵输入转换为多快拍平均的声压互谱矩阵输入，并基于互谱矩阵对角重构降噪建立了抗噪声干扰能力增强的EMAP-EM（Enhanced MAP-EM，EMAP-EM）压缩球波束形成方法。仿真和试验均表明，提出的MAP-EM和EMAP-EM压缩球波束形成均具有高的空间分辨率和计算效率，且EMAP-EM压缩球波束形成抗噪声干扰能力更强，尤其在低频、低信噪比环境中声源识别性能更佳。最后，分析了迭代次数和快拍数对MAP-EM和EMAP-EM压缩球波束形成性能的影响规律并给出推荐值。

关键词: 声源识别, 球传声器阵列, 压缩球波束形成, 稀疏贝叶斯学习

Abstract: Compressive spherical beamforming(CSB) based on solid spherical microphone array enjoys the advantages of panoramic sound imaging, suitable for medium and long distance measurement and easy to arrange, so it has broad application prospects in the field of noise source identification such as automobiles and airplanes. The recently proposed CSB based on Maximum type-II likelihood estimation(MLE-II) under sparse Bayesian learning(SBL) framework can achieve good low-frequency sound source identification performance, but it needs to estimate the sound source sparsity, and the anti-noise interference ability and computational efficiency also need to be improved. To solve the above problems, MAP-EM-based CSB(MAP-EM-CSB) without sound source sparsity estimation is proposed, which transforms the mathematical model solving problem of CSB into the maximum a posteriori(MAP) estimation problem of the source intensity distribution under SBL framework, and then uses the expectation maximization(EM) algorithm to solve it. Further, the sound pressure complex matrix input of MAP-EM-CSB is converted into the average sound pressure cross-spectrum matrix input of multiple snapshots, and then the enhanced MAP-EM-CSB(EMAP-EM-CSB) is established based on the diagonal reconstruction and noise reduction of the cross-spectrum matrix. Both simulations and experiment show that the proposed MAP-EM-CSB and EMAP-EM-CSB have high spatial resolution and computational efficiency. Due to its strong anti-noise interference capability, EMAP-EM-CSB has better sound source identification performance, especially in low frequency and low signal-to-noise ratio environments. Finally, the effects of the number of iterations and snapshots on the performance of MAP-EM-CSB and EMAP-EM-CSB are analyzed, and the recommended values are obtained.

Key words: sound source identification, spherical microphone array, compressive spherical beamforming, sparse Bayesian learning

中图分类号:

TB52

褚志刚, 赵洋, 杨亮, 张晋源, 杨洋. 基于稀疏贝叶斯学习的压缩球波束形成声源识别方法[J]. 机械工程学报, 2023, 59(2): 1-13.

CHU Zhigang, ZHAO Yang, YANG Liang, ZHANG Jinyuan, YANG Yang. Compressive Spherical Beamforming Based on Sparse Bayesian Learning for Sound Source Identification[J]. Journal of Mechanical Engineering, 2023, 59(2): 1-13.

参考文献

[1] 褚志刚, 杨洋, 贺岩松, 等. 球谐函数波束形成声源识别扩展方法[J]. 机械工程学报, 2015, 51(20):45-53. CHU Zhigang, YANG Yang, HE Yansong, et al. Extension method of spherical harmonics beamforming for sound source identification[J]. Journal of Mechanical Engineering, 2015, 51(20):45-53.
[2] 李敏宗, 卢奂采, 金江明. 用刚性表面球形传声器阵列重构入射波声场与识别定位声源[J]. 传感技术学报, 2015, 28(10):1459-1466. LI Mingzong, LU Huancai, JIN Jiangming. Reconstruction of incident sound field and identification of sound source using a rigid spherical microphone array[J]. Chinese Journal of Sensors and Actuators, 2015, 28(10):1459-1466.
[3] CHU Z, ZHAO S, YANG Y, et al. Deconvolution using CLEAN-SC for acoustic source identification with spherical microphone arrays[J]. Journal of Sound and Vibration, 2019, 440:161-173.
[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] RAFAELY B. Plane-wave decomposition of the sound field on a sphere by spherical convolution[J]. Journal of the Acoustical Society of America, 2004, 116(4):2149-2157.
[6] 褚志刚, 陈涛, 赵书艺, 等. 基于声压贡献的球面阵波束形成声源识别滤波求和算法[J]. 机械工程学报, 2018, 54(4):238-244. CHU Zhigang, CHEN Tao, ZHAO Shuyi, et al. Filter-and-sum beamforming sound source identification algorithm for spherical microphone arrays based on pressure contrition[J]. Journal of Mechanical Engineering, 2018, 54(4):238-244.
[7] YANG Y, CHU Z, SHEN L, et al. Functional delay and sum beamforming for three-dimensional acoustic source identification with solid spherical arrays[J]. Journal of Sound and Vibration, 2016, 373:340-359.
[8] 杨洋, 褚志刚. 高性能波束形成声源识别方法综述[J]. 机械工程学报, 2021, 57(24):166-183. YANG Yang, CHU Zhigang. A review of high-performance beamforming methods for acoustic source identification[J]. Journal of Mechanical Engineering, 2021, 57(24):166-183.
[9] YIN S, CHU Z, ZHANG Y, et al. Adaptive reweighting homotopy algorithm based compressive spherical beamforming with spherical microphone arrays[J]. Journal of the Acoustical Society of America, 2020, 147(1):480-489.
[10] 杨咏馨, 褚志刚, 杨洋. 基于正交匹配追踪的二维离网压缩波束形成声源识别方法[J]. 机械工程学报, 2022, 58(11):88-97. YANG Yongxin, CHU Zhigang, YANG Yang. Two-dimensional off-grid compressive beamforming based on orthogonal matching pursuit for acoustic source identification[J]. Journal of Mechanical Engineering, 2022, 58(11):88-97.
[11] TROPP J A, GILBERT A C. Signal recovery from random measurements via orthogonal matching pursuit[J]. IEEE Transactions on Information Theory, 2007, 53(12):4655-4666.
[12] 宁方立, 卫金刚, 刘勇, 等. 压缩感知声源定位方法研究[J]. 机械工程学报, 2016, 52(19):42-52. NING Fangli, WEI Jingang, LIU Yong, et al. Study on sound sources localization using compressive sensing[J]. Journal of Mechanical Engineering, 2016, 52(19):42-52.
[13] WIPF D P, NAGARAJAN S. Iterative reweighted l₁ and l₂ methods for finding sparse solutions[J]. IEEE Journal of Selected Topics in Signal Processing, 2010, 4(2):317-329.
[14] ZHANG S, LIU Y, LI X, et al. Variational Bayesian sparse signal recovery with LSM prior[J]. IEEE Access, 2017, 5:26690-26702.
[15] WIPF D P, RAO B D. An empirical Bayesian strategy for solving the simultaneous sparse approximation problem[J]. IEEE Transactions on Signal Processing, 2007, 55(7):3704-3716.
[16] XENAKI A, BOLDT J B, CHRISTENSEN M G. Sound source localization and speech enhancement with sparse Bayesian learning beamforming[J]. Journal of the Acoustical Society of America, 2018, 143(6):3912-3921.
[17] HUANG K, YANG Z. Robust sparse Bayesian learning for sparse signal recovery under unknown noise distribution[J]. Circuits Systems and Signal Processing, 2020, 40(3):1365-1382.
[18] GERSTOFT P, XENAKI A. Multisnapshot sparse Bayesian learning for DOA[J]. IEEE Signal Processing Letters, 2016, 23(10):1469-1473.
[19] GERSTOFT P, NANNURU S, MECKLENBRAUKER C F, et al. DOA estimation in heteroscedastic noise[J]. Signal Processing, 2019, 161:63-73.
[20] CHEN L, BI D, PAN J. Two-dimensional angle estimation of two-parallel nested arrays based on sparse Bayesian estimation[J]. Sensors, 2018, 18(10):3553.
[21] NANNURU S, GERSTOFT P, PING G, et al. Sparse planar arrays for azimuth and elevation using experimental data[J]. Journal of the Acoustical Society of America, 2021, 149(1):167-178.
[22] PING G, FERNANDEZ-GRANDE E, GERSTOFT P, et al. Three-dimensional source localization using sparse Bayesian learning on a spherical microphone array[J]. Journal of the Acoustical Society of America, 2020, 147(6):3895-3904.
[23] BISHOP C. Pattern recognition and machine learning[M]. New York:Springer, 2006.
[24] TIPPING M. Sparse Bayesian learning and the relevance vector machine[J]. Journal of Machine Learning Research, 2001, 1(3):211-244.
[25] DINSENMEYER A, ANTONI J, LECLERE Q, et al. A probabilistic approach for cross-spectral matrix denoising:Benchmarking with some recent methods[J]. Journal of the Acoustical Society of America, 2020, 147(2):3108-3123.

基于稀疏贝叶斯学习的压缩球波束形成声源识别方法

Compressive Spherical Beamforming Based on Sparse Bayesian Learning for Sound Source Identification

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	杨洋, 褚志刚, 杨咏馨. 典型因素对无网格压缩波束形成声源识别的影响[J]. 机械工程学报, 2022, 58(5): 98-107.
[2]	徐亮, 权璐纯, 尚俊超, 李敬豪, 张小正. 基于源强先验引导的正交匹配追踪声源识别算法[J]. 机械工程学报, 2022, 58(24): 20-31.
[3]	杨咏馨, 褚志刚, 杨洋. 基于正交匹配追踪的二维离网压缩波束形成声源识别方法[J]. 机械工程学报, 2022, 58(11): 88-97.
[4]	杨洋, 褚志刚. 高性能波束形成声源识别方法研究综述[J]. 机械工程学报, 2021, 57(24): 166-183.
[5]	柳明, 张捷, 高阳, 蒋文杰, 肖新标. 隧道内高速列车车内噪声特性及声源识别试验分析[J]. 机械工程学报, 2020, 56(8): 207-215.
[6]	樊小鹏, 余立超, 褚志刚, 杨洋, 李丽. 二维动态网格压缩波束形成声源识别方法[J]. 机械工程学报, 2020, 56(22): 46-55.
[7]	杨妍, 张捷, 何宾, 范静, 王德威, 肖新标. 基于试验测试的桥梁与路堤区段高速列车车外噪声特性分析[J]. 机械工程学报, 2019, 55(20): 188-197.
[8]	张捷, 肖新标, 姚丹, 伏蓉, 张骏. 基于心理声学参数的新型卧铺动车组包间噪声分析[J]. 机械工程学报, 2018, 54(4): 222-230.
[9]	褚志刚, 陈涛, 赵书艺, 肖新标. 基于声压贡献的球面阵波束形成声源识别滤波求和算法[J]. 机械工程学报, 2018, 54(4): 238-244.
[10]	徐亮, 胡鹏, 张永斌, 张小正. 可用于相干声源的快速反卷积声源成像算法[J]. 机械工程学报, 2018, 54(23): 82-92.
[11]	褚志刚, 段云炀, 沈林邦, 杨洋. 函数波束形成声源识别性能分析及应用^*[J]. 机械工程学报, 2017, 53(4): 67-76.
[12]	黎术, 徐中明, 贺岩松, 张志飞, 陈思. 基于函数广义逆波束形成的声源识别[J]. 机械工程学报, 2016, 52(4): 1-6.
[13]	褚志刚, 杨洋, 贺岩松, 康润程, 王光建. 球谐函数波束形成声源识别扩展方法[J]. 机械工程学报, 2015, 51(20): 45-53.
[14]	李兵;杨殿阁;郑四发;李克强;连小珉. 基于遗传算法的动态优化波叠加噪声源识别方法[J]. , 2010, 46(12): 99-105.
[15]	徐荣武;何琳;章林柯;贲可荣. 小样本条件下潜艇机械噪声源的识别[J]. , 2008, 44(7): 151-160.