确定相干机器多噪声源位置的方法研究

RESEARCH ON THE LOCATION OF COHERENT MECHANICAL NOISE SOURCES

提出利用均匀线阵接收的数据构造一种矩阵进行奇异值（ＳＶＤ）分解，以实现对相干噪声源的方位估计。利用这种方法，通过改变阵列中心位置便可确定机器多噪声源位置。计算机模拟和声学实验证实了这种方法的可行性。

The problem how to locate multiple coherent spherical wavefront noise sources with narrow band is solved in the paper. Based on the theory of array signal processing, a method is proposed to estimate bearings of multiple coherent mechanical noise sources with the singular value decomposition(SVD) of a data matrix and orthogonality of spectrum using a uniform linear array. Based on it, a method can be applied to locate the mechanical noise sources by changing the position of the array center using geometry relation of array. The theory of singular value decomposition(SVD) algorithms is verified by computer numerical tests and acoustics experiments in an anechoic room. The results of tests and experiments show that mean square error(MSE) of bearings and locations of multiple noise sources is little. Compared with typical method of multiple signal classification(MUSIC) based on the theory of array signal processing, the method has high resolution. It shows that the method is effective.