交叉层波束形成方法的发动机噪声源识别

发动机噪声源分布复杂，来源多，用人耳很难分辨，利用传声器阵列的噪声源识别技术可以为发动机噪声控制提供客观依据和指导。使用波束形成声源识别方法，对位于不同平面的多个声源进行了仿真识别，并研究了多维声源识别方法，使用交叉层法得到了声源定位的立体结果。结果显示，交叉层法可以有效消减或去除来自识别表面之外的声源在识别表面的虚假投影。最后，针对某发动机产品，使用平面传声器阵列对其上、前、左、右四个面分别进行一次变转速工况时域声压信号采集，使用互谱矩阵波束形成算法，得到各转速下发动机各表面的声源分布图像，并通过交叉层法得到了发动机表面声源的立体分布，准确将声源定位至发动机表面各部件。

Identification of engine noise source by beamforming based cross-layer method

The noise sources of engine are very complex and difficult to be identified by ear. The noise source identification technique using microphone array can provide an objective basis and guidance for engine noise control. Based on beamforming technique, the multi-dimensional source recognition method is studied and the recognition of multiple sources in different planes is simulated. The cross-layer method can get the stereo results of source positioning, and it can effectively reduce or remove the false projections on the recognition surface from the sources outside the recognition surface. Finally, for a certain engine product, the plane microphone array is used to collect the sound pressure signals from the top, front, left and right sides of the engine at different speeds, and the cross-spectrum matrix beamforming algorithm is used to obtain the sound source distribution image on each engine surface and at each speed. By the cross-layer method, the stereo distribution of the sound sources on the engine surface is obtained, and the sound sources are accurately located to the corresponding parts of the engine surface.