基于正交最小二乘的声源高分辨识别定位方法研究
Research on high-resolution sound source recognition localization method based on orthogonal least square
丁林宁 1魏明洋 2康雅聪1
作者信息
- 1. 南京理工大学电子工程与光电技术学院,江苏南京 210094
- 2. 中国科学院声学研究所,北京 100190
- 折叠
摘要
针对小孔径平面阵下的低频相干源定位分辨率低的问题,提出一种基于正交最小二乘(OLS)的协方差拟合定位法.首先,将OLS运用到协方差拟合成本函数中;然后,利用OLS能在迭代识别源时能逐渐减少点扩散函数(PSF)分量影响的特点,在每次迭代后,利用当前源信息重新校正先前识别源,并通过指数参数提高声源选择标准以缩减各声源主瓣宽度提高空间分辨率.仿真和试验表明,所提方法显著降低了声源间距固定的双相干声源可分辨的最低频率.
Abstract
Aiming at the problem of low resolution of low frequency coherent source localization in small aperture planar array,a covariance fitting localization method based on orthogonal least square(OLS)is proposed.Firstly,the OLS is applied to the covariance fitting function,and then when the OLS is used to identify the source iteratively,the influence of the point spread function(PSF)component can be gradually reduced.After each iteration,the previous identified sound sources are re-corrected using the current sound source information,and the selection criteria of sound sources are improved through the exponential parameters to reduce the main lobe width of each sound source and improve the spatial resolution.Simulation and experiment show that the proposed method significantly reduces the lowest frequency of the double coherent sound source with fixed sound source spacing.
关键词
传声器阵列/声源定位/正交最小二乘/协方差矩阵Key words
sensor arrays/sound source localization/orthogonal least square(OLS)/covariance matrix引用本文复制引用
基金项目
中科院声学所横向技术类项目(492021001994)
出版年
2024
NETL
NSTL
万方数据