This paper presents an alternative energy function for Global Optimization (GO) beamforming, tailored to acoustic broadband sources. Given, that properties such as the source location, multipole rotation, or flow conditions are parameterized over the frequency, a CSM-fitting can be performed for all frequencies at once. A numerical analysis shows that the nonlinear energy function for the standard GO problem is equivalent to the source's Point Spread Function (PSF) and contains local minima at the grating- and side lobes' locations. The energy function is improved with the proposed broadband energy, as it averages the PSF. Further, it simplifies the process of identifying sources and reconstructing their spectra from the results. The paper shows that the method is superior on synthetic monopoles compared to standard GO and CLEAN-SC. For real-world data the results of the proposed method and CLEAN-SC are similar, and outperform standard GO. The main difference is that source assumption violations cause noisy maps for CLEAN-SC and cause wrong spectral estimations of the proposed method. By using reasonable initial values, the GO problem reduces to a Local Optimization problem with similar results. Further, the proposed method is able to identify synthetic multipoles with different pole amplitudes and unknown pole rotations.

翻译：本文提出了一种适用于声学宽带声源的全局优化波束形成替代能量函数。通过将声源位置、多极子旋转或流动条件等属性参数化为频率的函数，可一次性对所有频率进行互谱矩阵拟合。数值分析表明，标准全局优化问题的非线性能量函数等效于声源的点扩散函数，并在栅瓣和旁瓣位置存在局部极小值。所提出的宽带能量函数通过平均点扩散函数改进了原有能量函数，同时简化了从结果中识别声源及重构其频谱的过程。本文证明，该方法在合成单极子声源上优于标准全局优化和CLEAN-SC算法。对于实际数据，所提方法与CLEAN-SC的结果相似，且均优于标准全局优化算法。主要区别在于：当声源假设不满足时，CLEAN-SC会产生噪声地图，而所提方法则会导致频谱估计错误。通过采用合理的初始值，全局优化问题可简化为具有相似结果的局部优化问题。此外，所提方法能够识别具有不同极子幅度和未知极子旋转的合成多极子声源。