摘要
为了克服传统波束形成声源识别方法受旁瓣限制而无法识别次声源的缺点,将奇异值分析与波束形成算法相结合,给出了奇异值分解波束形成声源识别新方法:首先对阵列传声器测量信号的互谱矩阵进行奇异值分解重构,然后基于分解后的各互谱矩阵进行波束形成运算,从而识别主声源及各次声源。在仿真及算例试验均验证了其准确性及有效性的基础上,利用奇异值分解波束形成对某汽车前围板隔声薄弱环节进行识别,结果表明:空调进风口为第一声源,暖风机进出水管安装孔及空调漏水管安装孔分别为第二和第三声源。该方法准确高效,能够获得更全面的声源信息,提高了波束形成的声源识别性能,为识别次声源提供了有力工具。
In order to overcome the disadvantage that the minor source cannot be identified because of the side lobe limitation by traditional beamforming method, a new sound source identification method named singular value de-composition beamforming was presented, which combined the singular value decomposition and beam-forming algo-rithm.Signal cross-spectral matrix measured by microphones was decomposed and recombined, and then the main source and other minor sources can be identified when the decomposed cross-spectral matrix was calculated with beam-forming.The simulation and verifiable experiment validated the correctness and effectiveness of the new method.The weak parts of a vehicle’ s dash panel were identified by singular value decomposition beam-forming. The results show that the first source is the air condition inlet, the installation holes of air blower outlet-inlet pipe and the leak pipe of air condition are the second and third source, respectively.The method can improve the per-formance of sound source identification and provides a powerful tool for the identification of minor sources for its ac-curacy and efficiency in getting more comprehensive information of sound source.
出处
《电子测量与仪器学报》
CSCD
2014年第11期1177-1184,共8页
Journal of Electronic Measurement and Instrumentation
基金
国家自然科学基金(50975296)
中央高校基本科研业务费(CDJZR13110001)资助项目
关键词
声源识别
波束形成
奇异值分解
次声源
试验
sound source identification
beamforming
singular value decomposition
minor source
experiment
