Page 115 - 应用声学2019年第2期

第 38 卷 第 2 期 Vol. 38, No. 2
2019 年 3 月 Journal of Applied Acoustics March, 2019

⋄ 研究报告 ⋄

基于Group Lasso的多重信号分类声源

定位优化算法 ∗

吴江涛 胡定玉 † 方 宇 朱文发

(上海工程技术大学城市轨道交通学院 上海 201620)
摘要 多重信号分类算法因其抑制噪声能力强、计算速度快等优点，在声源定位领域得到广泛应用。但该算法
在中低频段分辨率及聚焦性能较差。针对该问题，提出一种基于 Group Lasso 的多重信号分类优化算法。该算
法将多重信号分类算法输出值作为初始值，并在 Group Lasso 算法组间计算时对目标信号进行稀疏、在组内计
算时对该组信号进行平滑及阈值截断。仿真结果表明：该优化算法在中低频段可明显提高多重信号分类算法
分辨率，同时改善因扫描位置与声源面位置不重合引起的聚焦性能下降问题。
关键词 多重信号分类算法，Group Lasso，声源定位
中图法分类号: TB52+5 文献标识码: A 文章编号: 1000-310X(2019)02-0261-06
DOI: 10.11684/j.issn.1000-310X.2019.02.016

An optimized multiple signal classification algorithm based on

Group Lasso for sound localization

WU Jiangtao HU Dingyu FANG Yu ZHU Wenfa

(School of Urban Rail Transportation, Shanghai University of Engineering Science, Shanghai 201620, China)

Abstract Multiple signal classification (MUSIC) algorithm is widely used in the field of sound source local-
ization due to its robustness to noise and computation efficiency. However, this algorithm has poor resolution
and focusing performance in the low and medium frequency bands. In view of this problem, a MUSIC algorithm
optimized by Group Lasso algorithm is proposed. The output of MUSIC algorithm is used as the initial value.
When the Group Lasso algorithm group is calculated, the target signal is sparse and calculated in the group.
The set of signals is smoothed and the threshold is truncated. The simulation results show that the optimized
algorithm can significantly improve the resolution of the MUSIC algorithm in the middle and low frequency
bands, and at the same time, the problem of degraded focusing performance caused by the non-coincidence of
scanning position and sound source surface position is improved.
Key words Multiple signal classification algorithm, Group Lasso, Source location

2018-05-21 收稿; 2018-10-15 定稿
国家自然科学基金青年基金项目 (51605274), 上海工程技术大学展翅计划项目 (RC152017), 上海工程技术大学研究生科研创新项目
∗
(17KY1012)
作者简介: 吴江涛 (1993- ), 男, 山东潍坊人, 硕士研究生, 研究方向: 噪声源识别与定位。
† 通讯作者 E-mail: dyhu1987@sues.edu.cn