非均匀噪声下基于稀疏重构的声矢量传感器阵列方位估计方法  

A method of DOA estimation of acoustic vector sensor array based on sparse reconstruction under non-uniform noise

作　　者：王伟东 李向水 李辉 史文涛[2] WANG Weidong;LI Xiangshui;LI Hui;SHI Wentao(College of Physics and Electronic Information Engineering,Henan Polytechnic University,Jiaozuo 454000,China;School of Marine Science and Technology,Northwestern Polytechnic University,Xi’an 710072,China)

机构地区：[1]河南理工大学物理与电子信息学院,河南焦作454000 [2]西北工业大学航海学院,西安710072

出　　处：《振动与冲击》2023年第13期127-136,共10页Journal of Vibration and Shock

基　　金：国家自然科学基金(62101176);河南省高等学校重点科研项目计划支持(22A510006);河南理工大学博士基金资助项目(B2022-3);河南省重点研发与推广专项(科技攻关)支持(232102211004)。

摘　　要：为解决非均匀噪声情况下声矢量传感器阵列方位估计性能恶化的问题。提出基于加权最小二乘(weighted least squares,WLS)的稀疏信号重构法和基于加权协方差矩阵拟合(weighted covariance matrix fitting,WCMF)的稀疏信号重构法。定义一个虚拟的声矢量传感器阵列流形矩阵,并重构包含稀疏信号功率和噪声功率的协方差矩阵。为估计稀疏信号功率和每个通道输出的噪声功率,基于WLS法和稀疏信号加权最小化法,构造了关于稀疏信号功率和噪声功率的代价函数。在此基础上,为了进一步提高稀疏信号功率和噪声功率的估计精度,基于WCMF准则对构造的代价函数进行改进。应用泰勒级数展开式将关于稀疏信号功率和噪声功率的非线性代价函数转化为线性函数,并采用循环迭代算法估计稀疏信号功率和噪声功率,待迭代终止时,对稀疏信号功率谱峰搜索,即可实现对目标的方位估计。仿真结果表明,与现有非均匀噪声下的估计方法相比,所提方法提高了非均匀噪声情况下声矢量传感器阵列的方位估计精度。Here,to solve the problem of direction of arrival(DOA)estimation performance of acoustic vector sensor array deteriorating under non-uniform noise,sparse signal reconstruction methods based on weighted least squares(WLS)and weighted covariance matrix fitting(WCMF),respectively were proposed.Firstly,a virtual manifold matrix of acoustic vector sensor array was defined,and a covariance matrix containing sparse signal power and noise power was reconstructed.Then,in order to estimate sparse signal power and noise output power of each channel,the cost function of sparse signal power and noise power was constructed based on WLS and the sparse signal weighted minimization method.Furthermore,in order to further improve estimation accuracy of sparse signal power and noise power,the constructed cost function was improved based on WCMF criterion.Finally,the nonlinear cost function of sparse signal power and noise power wasconverted into a linear function by using Taylor series expansion,and the cyclic iterative algorithm was used to estimate sparse signal power and noise power.When the iteration stopped,searching power spectrum peak of sparse signal could realize DOA estimation of targets.The simulation results showed that compared with existing estimation methods under non-uniform noise,the proposed methodscan improve the DOA estimation accuracy of acoustic vector sensor array under non-uniform noise.

关 键 词：声矢量传感器阵列 非均匀噪声 稀疏重构 方位估计 

分 类 号：TG911.7[金属学及工艺—钳工工艺]