双面声阵列波束形成的正则化改进算法

发布时间：2018-03-12 20:08

本文选题：声阵　切入点：正则化方法　出处：《声学学报》2017年02期 　论文类型：期刊论文

【摘要】：双面声阵列波束形成能够区分识别位于不同扫描平面的相干声源,然而该算法在低信噪比条件下识别精度较低。针对此问题提出一种迭代正则化改进算法,通过迭代方法更新正则化矩阵与波束形成输出,在不断提升正则化稳定性、抑制干扰旁瓣的基础上使声学云图主瓣向实际相干声源点处聚焦。数值仿真与实验算例结果显示,改进算法在中高频代表频率下能够正确区分相干声源前后方位,并具有相对原算法更高的识别精度。从而表明:从反问题正则化角度对原算法进行优化改进是理论可行的;正则化矩阵的具体形式与广义逆波束形成输出的空间分辨率紧密相关,且可通过迭代方法将二者整合以提高声源识别精度。
[Abstract]:Dual acoustic array beamforming can distinguish coherent sound sources located in different scanning planes, but the accuracy of the algorithm is low under low SNR. An improved iterative regularization algorithm is proposed to solve this problem. The regularization matrix and beamforming output are updated by iterative method. On the basis of increasing the regularization stability and suppressing the interference sidelobes, the main lobe of the acoustic cloud image is focused towards the actual coherent sound source point. The numerical simulation and experimental results show that, The improved algorithm can correctly distinguish the front and rear azimuth of coherent sound source at the representative frequency of middle and high frequency, and has higher recognition accuracy than the original algorithm, which shows that it is feasible to optimize the original algorithm from the perspective of inverse problem regularization. The specific form of the regularization matrix is closely related to the spatial resolution of the generalized inverse beamforming output, and the two can be integrated by iterative method to improve the accuracy of acoustic source identification.
【作者单位】：重庆大学机械传动国家重点实验室;重庆大学汽车工程学院;
【基金】：国家自然科学基金项目(51275540) 重庆市重大应用开发计划项目(cstc2015yykfC60003)资助
【分类号】：TN912.3

【相似文献】