分段取阈值正交匹配追踪反卷积声源识别算法  

Deconvolution sound soure indentification algorithm based on piecewise thresholding OMP

作　　者：赵卫鹏 毛锦 刘凯[1] 杜进辅[1] ZHAO Weipeng;MAO Jin;LIU Kai;DU Jinfu(School of Mechanical and Precision Instrument Engineering,Xi’an University of Technology,Xi’an 710048,China)

机构地区：[1]西安理工大学机械与精密仪器工程学院,西安710048

出　　处：《振动与冲击》2023年第15期268-276,共9页Journal of Vibration and Shock

基　　金：陕西省自然科学基础研究计划(2022JM271);陕西省博士后科研项目(2018BSHEDZZ10)。

摘　　要：正交匹配追踪反卷积声源识别算法(orthogonal matching pursuit deconvolution approach for the mapping of acoustic sources, OMP-DAMAS)具有极高的计算效率、空间分辨率和重构精度,被广泛应用声源识别领域。但在实际的工程运用中,无法满足提前确定声源数目的条件,可能造成识别结果不准确。因此提出了一种分段取阈值的OMP-DAMAS算法,在声源稀疏度未知的情况下,通过对内积和最小二乘解取阈值将伪声源和旁瓣对应的列序号从原子支撑集中删除,直接精确的识别出真实声源的位置。仿真和试验结果表明了所提算法与传统的延时求和算法相比,可以明显的减小主瓣宽度,提高空间分辨率,同样能达到OMP-DAMAS算法的重构效果,对噪声具有较好的鲁棒性,且具有极高的识别稳定性。Orthogonal matching pursuit deconvolution sound soure indentification algorithm(OMP-DASRA)is widely used in the field of acoustic source recognition due to its higher computational efficiency,spatial resolution and reconstruction accuracy.However,in practical engineering applications,due to conditions to determine number of sound sources in advance being unable to meet,the algorithm may cause incorrect recognition results.Here,an OMP-DASRA with piecewise thresholding was proposed.When the sparsity of sound sources was unknown,column sequence numbers corresponding to pseudo sound sources and side lobes were deleted from atomic support set by taking threshold values for inner product and least squares solutions to directly and accurately recognize positions of real sound sources.Simulation and experimental results showed that compared to traditional time-delay summation algorithms,the proposed algorithm can obviously reduce main lobe width,improve spatial resolution,and reach the reconstruction effect of OMP-DASRA;it has better robustness to noise,and higher recognition stability.

关 键 词：正交匹配追踪 反卷积 声源识别算法 阈值 

分 类 号：TB52[理学—物理]