基于相干分析的波束形成算法研究

打开文本图片集

中图分类号：TB52；TN911.7文献标志码：A doi：10.3969/j.issn.1006-0316.2026.02.005

文章编号：1006-0316（2026）02-0032-07

StudyofCoherence-based Acoustic BeamformingAlgorithms

TANG Yanning'，GUO Weiqiang’，LI Xinyi1，ZHAO Jingzhou1， WANG Zeyang²，LAN Xiaodan³，WANG Di4

（1.CRRC Changchun Railway Vehicles Co.，Ltd.， Changchun 130062，China; 2.Institute of Smart City and Intelligent Transportation，Southwest JiaoTong University， Chengdu 61o031， China; 3.School of Mechanical Engineering，Southwest JiaoTong University，Chengdu 61oo31，China; 4.Siemens Industry Software （Beijing） Co.， Ltd.， Beijing， China）

Abstract ∵ Coherent output spectral analysis （COSA） and conditioned spectral analysis （CSA） are typical signal separation techniques.Compared with COSA， which is limited to onlyone reference signal， CSA can be applied to multi-reference signals，which offers broader engineering applicability.The array microphone signals and the reference signals of excitation sources are measured simultaneously，and the results of the COSA and CSA performed on the arry microphone signals are applied on preprocessing of the acoustic beamforming algorithm. Acoustic beamforming algorithms based on COSA and CSA are then proposed separately. The efficacy of the proposed methods is examined experimentally using a two-dimensional circular microphone array to locate three artificial white noise sources.Firstly，the results of COSA-basedacoustic beamforming algorithmare compared to the results of conventional algorithm，which verifies the efcacy of COSA-based acoustic beamforming algorithm in separating or suppressing the sound sources in the acoustic map.Secondly，the resultsof CSA-based acoustic beamforming algorithm using multi-reference signalsare compared to the results of conventional beamformingalgorithm，whichverifies the eficacyofCSA-based acoustic beamforming algorithm in sound source saparation and the applicability to multi-reference systems.

Key words ：acoustic beamforming ;coherent output spectrum ;conditioned spectrum multi-reference model ; sound source separation

传统波束形成算法可以进行高效的声源定位，并通常以声源云图的方式显示总的声源位置和声源强度信息[1-3]，该信息包括了所有激励源产生的声源的叠加。（剩余6575字）