未知噪声背景下多声源方位角的估计方法

以平面波、窄带声源的阵列测向模型为基础,利用阵列信号处理的理论和方法,在假设理想的白噪声背景下,对均匀线阵接收的信号数据,采用多信号分类法（ＭＵＳＩＣ法）,利用其正交性的谱估计来判断声源的方位角。在此基础上,对不少实际环境中的噪声模型是非白和不确定的情况下,提出一种辅助阵元的方法,以确定在未知噪声背景下利用ＭＵＳＩＣ法确定多声源的方位角。声学实验证实,在实际环境中,该方法能很好地分辨多声源的方位角,且标准差很小。通过与假想的白噪声环境进行对比,本文方法能有效地消除实际环境带来的测向误差,大大提高测向分辨率,由此证实了该方法的有效性。

A Method for Azimuth Angle Estimations of Sources in Unknown Noise Environments

Based on a array direction finding model of multiple planar wavefront sources with narrow band, using the theory and the method of array signal processing, a method of multiple signal classification (MUSIC) and orthogonality of spectrum is proposed to estimate azimuth angles of multiple sources using a uniform linear array in the background of ideal white noise. Based on it, in the circumstances of nonwhite and uncertain noise model, a method of additional array component is proposed to estimate azimuth angles of sources in unknown noise environment using MUSIC. The theory is verified by acoustics experiments in anechoic room. The experimental results show that the mean square error (MSE) of the azimuth angles of multiple sources is little in the real environment. Compared with ideal environment of white noise, the method has higher resolution and is proved to be effective.