混响背景下水底目标回波盲分离  

Blind separation of underwater target echoes in reverberation background

作　　者：李秀坤[1,2] 夏峙[1,2] 王向红[1,2] 牟林[3] 

机构地区：[1]哈尔滨工程大学水声技术重点实验室,黑龙江哈尔滨150001 [2]哈尔滨工程大学水声工程学院,黑龙江哈尔滨150001 [3]中国船舶重工集团公司第760研究所,辽宁大连116013

出　　处：《哈尔滨工程大学学报》2015年第1期62-67,共6页Journal of Harbin Engineering University

基　　金：国家自然科学基金资助项目(51279033)

摘　　要：针对水下掩埋目标探测中目标回波信混比较低的问题,提出将混响视为一类信源的与目标回波盲分离方法,研究了目标回波与混响在时域统计特性、时频域以及时域波形上的可分离性与分离原则问题,解决了传统时频方法不适用于当混响与目标回波在时频域内混叠时的抑制混响问题。海上掩埋目标探测实验数据处理结果表明,经过时域二阶统计盲分离与时频域盲分离处理后,目标回波的信混比提高了约5 d B,在时频分布上可以更清楚的观察到目标几何亮点,从而提高对目标回波识别的准确性。Considering the low signal-reverberation ratio（ SRR） of target echoes in underwater buried object detection,a method of separation between a target echo and reverberation is proposed. In this paper,the reverberation is regarded as a kind of source. The statistical characteristics of the target echo and reverberation in time domain and separation principle of waveform in time and time-frequency domains are researched. Addressing the problem related to the traditional signal processing method being inappropriate to suppress reverberation when target echo and reverberation is aliasing in time-frequency domain is solved. The processing results of buried object detection experiment on the sea showed that the SRR of the target echo increases by 5 d B after the processing by the blind source separation of two-order statistics in time domain and the blind source separation in time-frequency domain. The geometrical bright spot of the target can be observed clearly in time frequency distribution in which the accuracy of target echo recognition is improved.

关 键 词：水声探测 掩埋目标 混响 信号混叠 可分离性 信号分离 盲源分离 

分 类 号：TN911.7[电子电信—通信与信息系统]