主动声纳的时频盲分离算法研究  

Research on Time-Frequency Blind Separation Algorithm for Active Sonar

作　　者：杨阳[1,2] 张诚 丁元明[1,2] YANG Yang;ZHANG Cheng;DING Yuan-ming(Key Laboratory of Communication and Network,Dalian University,Dalian,Liaoning 116622,China;College of Information Engineering,Dalian University,Dalian Liaoning 116622,China)

机构地区：[1]大连大学通信与网络重点实验室,辽宁大连116622 [2]大连大学信息工程学院,辽宁大连116622

出　　处：《计算机仿真》2021年第5期193-198,共6页Computer Simulation

基　　金：国家自然科学基金(61901079)。

摘　　要：主动声纳目标探测中,水雷等浅海安静型小目标声散射回波受海洋环境干扰严重,影响对目标特征的准确提取与识别,探测可靠性低。针对上述问题,研究一种结合形态学滤波的时频域盲分离算法,根据主动声纳回波中目标与混响的时频特征差异,设计形态学结构元的几何结构,滤除与目标谱不重叠的混响及互项干扰,再利用波形聚类方法优化与目标谱重叠的混响干扰,突出目标亮点的时频谱线,得到目标声散射的自项时频信息,最后获得目标亮点的时频域解混矩阵,分离出目标声散射信号。数值仿真结果表明,结合形态学滤波的时频域盲分离算法,提高了目亮点信号的信混比。In active sonar target detection, the acoustic scattering echoes of shallow small quiet targets such as mines are seriously interfered by the marine environment, which affects the accurate extraction and recognition of target features, and the detection reliability is low. To solve this problem, a blind separation algorithm in the time-frequency domain combined with morphological filtering was studied. According to the difference between the time-frequency characteristics of the target and the reverberation in the active sonar echo, the geometric structure of the morphological structural elements was designed to filter out the target. Then, the waveform clustering method was used to optimize the reverberation interference overlapped with the target spectrum to highlight the time-frequency spectrum line of the target bright spot, obtain the target time-frequency information of the target sound scattering, and finally obtain the target bright spot. The time-frequency domain unmixing matrix separates the target sound scattering signal. The numerical simulation results show that the blind separation algorithm in time-frequency domain combined with morphological filtering improves the signal-to-mix ratio of the bright spot signal.

关 键 词：水下目标探测 混响抑制 盲分离 时频特性 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]