利用時域之近場等效聲源影像法進行聲源識別和聲場重建

Abstract

在研究計畫中提出麥克風陣列法(microphone array)來進行平面及非平面之噪音源診斷。而本計劃所提出的麥克風陣列架構主要是以時域之反逆濾波器(inverse filter)的基楚，進而發展出近場等效聲源影像(nearfield equivalence source imaging, NESI)技術來實現聲源識別和聲場重建，進而提升噪音源判斷的正確率。本計劃之NESI是在時域上設計，它除了可以應用在穩定噪音源上亦可應用在非穩定噪音源上。並且利用最小均方最佳化來設計出多通道反逆濾器。Tikhonov調整化被利用來仰制模型相稱的不好條件值。此外，可預期出的問題如：重建聲場表面上的奇異(singularity)問題可以使用撤退焦點表面法(retreated focus point)來解決、邊界上失焦的問題可以使用窗口矩陣的設計來解決等。最後使用高效率之狀態空間最小化的獲得技術(stat-space minimal realization technique)來實現多通道反逆濾器。而在聲學學變數上，NESI除了可以計算出聲音的強度(soruce strength)外，亦可進一步計算出重建面之聲壓(sound pressure)、粒子速度(particle velocity)及聲音密度(sound intensity)等。NESI除了可以應用在一維及二維矩形麥克風陣列以外，亦可應用在不規則之幾合形狀之麥克風陣列上。此外，NESI也可以去估測出遠場的聲壓大小。本計劃第一年度將以發展NESI的演算法及模擬為主要，第二年將以後處理及計算聲學變數為主，第三年的工作會將這些演算法、後處理及計算聲學變數用硬體實現。

The microphone array technology is proposed to identify locations and strengths of noise sources. The Nearfield Equivalence Source Imaging (NESI) is based on the time-domain formulation that applies not only to stationary but also a transient noise. Multichannel inverse filters are designed using the least-square optimization. Tikhonov regularization is called for to mitigate the ill-posedness inherent in the underdetermined model-matching problem. A windowing design is also suggested to cope with boundary defocusing problem. Beam pattern is calculated for the inverse filters with window design. In addition, a modified approach based on retreated focus points is devised to avoid the singularity problem in reconstructing sound images. The multichannel inverse filters are implemented in light of a highly efficient state-space minimal realization technique. Furthermore, acoustical variables are computed by the post-processing. The NESI is also use in farfield prediction to estimate sound pressure in the farfield. In the first stage of the project, the theory and simulation of NESI will be developed. In the second stage of the project, the post-processing will be developed to calculate acoustical variables. In the final stage of the project, the softwar and the hadware of NESI will be intergrated into a complete system.