规整化最小二乘子空间相交测向技术探讨  

Discussion on regularized least-square subspace intersection DOA estimation method

作　　者：巩玉振[1] 张秋生[1] 刘云涛 蔡惠智[1] 

机构地区：[1]中国科学院声学研究所,北京100190

出　　处：《声学技术》2013年第3期198-202,共5页Technical Acoustics

摘　　要：在浅海条件下,子空间相交的方法可以消除多途效应的影响,实现水平长阵的准确目标方位估计,获得比常规波束形成算法更优的性能。然而,该算法在数值计算稳定性上存在问题。在最小二乘子空间相交算法的基础上,提出一种Tikhonov规整化的最小二乘子空间相交算法,该算法在原解向量的基础上,施加了一个滤波因子,滤除了小特征值对应的解向量,使算法的数值计算稳定性大大提高,从而也提高了算法的检测性能。试验数据处理结果表明,该算法是有效的。Bearing estimation is always a hard work in shallow water due to multimode nature of acoustic propagation in an oceanic waveguide.A new least-square subspace intersection（LSSI） method with Tikhonov’s regularization is proposed.Subspace intersection method yields unbiased bearing estimates.But it involves an inverse problem with a highly ill-conditioned coefficient matrix.In order to compute stable solutions it is necessary to apply regularization method.It is shown that Tikhonov’s regularization method is suitable for this problem,which can improve the robustness of LSSI algorithm.Results of processing sea-trial real data are presented.

关 键 词：DOA估计 Tikhonov规整化 子空间相交 

分 类 号：TN911.7[电子电信—通信与信息系统]