• CN:11-2187/TH
  • ISSN:0577-6686

机械工程学报 ›› 2017, Vol. 53 ›› Issue (4): 67-76.doi: 10.3901/JME.2017.04.067

• 异种材料的焊接与连接 • 上一篇    下一篇

函数波束形成声源识别性能分析及应用*

褚志刚1,2, 段云炀1,2, 沈林邦1,2, 杨洋1,3   

  1. 1. 重庆大学机械传动国家重点实验室 重庆 400044;
    2. 重庆大学汽车工程学院 重庆 400044;
    3. 重庆工业职业技术学院车辆工程学院 重庆 401120
  • 出版日期:2017-02-20 发布日期:2017-02-20
  • 作者简介:

    褚志刚(通信作者),男,1978年出生,博士,副教授。主要研究方向为车辆系统动力学与控制、汽车振动噪声控制、计算机辅助测试理论与技术。

    E-mail:zgchu@cqu.edu.cn

  • 基金资助:
    * 国家自然科学基金(51275538)和重庆市重大应用开发计划(cstc2015 yykfc60003)资助项目; 20160203收到初稿,20161205收到修改稿;

Performance Analysis and Application of Functional Beamforming Sound Source Identification

CHU Zhigang1,2, DUAN Yunyang1,2, SHEN Linbang1,2, YANG Yang1,3   

  1. 1. The State Key Laboratory of Mechanical Transmission, Chongqing University, Chongqing 400044;
    2. School of Automotive Engineering, Chongqing University, Chongqing 400044;
    3. Faculty of Vehicle Engineering, Chongqing Industry Polytechnic College, Chongqing 401120
  • Online:2017-02-20 Published:2017-02-20

摘要:

函数波束形成声源识别算法能够有效克服传统波束形成存在的“主瓣”宽、“旁瓣”高且多等缺陷。基于传统波束形成及矩阵特征值分解理论,给出函数波束形成算法的理论推导。基于自主开发的算法程序进行单声源、不相干双声源的模拟仿真,结果表明:函数波束形成在识别单声源、不相干声源时均能有效地衰减旁瓣,提高识别分辨率,且能清晰地识别弱源;当声源不在聚焦网格点时,函数波束形成主瓣峰值输出随指数的增大而降低,出现失真的情况;传声器及测量通道频响失配情况对函数波束形成的成像影响很小。由函数波束形成峰值输出误差随指数的变化曲线,建议指数一般取16。进一步,不相干四声源的算例试验验证了算法的有效性及程序的正确性。对某汽车前围板的隔声薄弱环节识别中,除空调进风口及暖风机进出水管安装孔隔声薄弱环节以外,亦准确定位转向柱安装孔附近这一隔声薄弱位置,识别结果更全面,为进一步改进其隔声性能指明了方向。

关键词: 波束形成, 互谱, 试验, 特征值分解, 声源识别

Abstract:

Functional beamforming algorithm can overcome the defects of wide main lobe and high sidelobes effectively, which appears in conventional beamforming. Based on conventional beamforming algorithm and the matrix eigenvalue decomposition theory, the theoretical derivation of function beamforming algorithm is presented. Imaging maps of the given single source and incoherent sources are simulated based on the algorithmic program self-development. The results are listed as following. For single source or incoherent sources, functional beamforming could not only suppress sidelobes effectively and improve resolution slightly but also identify the weak source clearly. Functional beamforming output of main lobe will reduce even appear the distortion with the growth of exponent. Frequency response mismatch between microphones and channels of array has less impact for the functional beamforming imaging. The exponent is suggested about 16 on the basis of the beamforming outputs error curves with exponent. Furthermore, the sound source identification experiments on four incoherent sounds were carried out, which verifies the effectiveness of this method and the correctness of the developed program. The weak part of car dash panel are identified by functional beamforming. Besides the air condition inlet and the installation holes of air blower outlet-inlet pipe, the mounting hole of steering column is also identified. The method gets more comprehensive information of sound source and provides direction for the improvement of sound insulation performance.

Key words: beamforming, cross-spectra, eigenvalue decomposition, test, sound source identification