基于阵列模型的窄带信号盲解卷积方法

Method of Narrowband Signal Blind Deconvolution Based on Array Model

在水声技术中,鱼雷对多目标进行被动检测时,需要对多目标源信号进行分离、定位,而雷体传感器阵列接收的信号通常为纯时延卷积混合信号。本文针对这种卷积混合只考虑窄带信号情况,把盲分离与阵列信号处理结合起来,将实数的卷积混合转化为复数的瞬时混合,从而在盲分离阵列模型的基础上,通过复数盲分离的手段完成盲解卷积。解卷积恢复的多目标源信号十分有利于多目标特征识别与定位。文中所构建的正弦信号盲解卷积仿真试验取得了令人满意的结果,表明方法是正确有效的。

In marine engineering technology, when targets are detected by a torpedo, we need to separate and locate the signals of targets. The signals that the torpedo sensor-array receives are always convolution mixture ones. Considering only the narrowband signal in the pure time-delayed convolution mixture, we combine blind separation with array signal processing, and transform real number convolution mixture to complex number instantaneous mixture. Therefore, based on array model of blind separation, we complete the blind deconvolution via complex number blind separation. Simulation results prove that the method is efficient, and the signals of multi-target source from sine signal deconvolution facilitate the feature identification and localization of multi-target.