基于盲解卷积的水声信号恢复技术

An underwater signal recovery technique based on blind deconvolution

信号在环境复杂多途严重的水声波导中传输后,接收到的信号时间长度被拉长,信号是失真的。在许多实际应用中,常常希望从已失真的接收信号中把原始信号波形恢复出来。本文利用盲解卷积技术对水声信号恢复进行研究。理论推演表明,在垂直阵条件下用人造的格林函数可成功地代替水声信道真实的格林函数,可以把非线性关系化解为线性关系,从而推导出依靠基阵记录下的信息去确定声源宽带信号原始波形和环境传播特征的公式和步骤。声场数值计算对6种海底类型进行,恢复后的宽带信号与原始信号的归一化相关系数均大于0.945,对硬海底多途严重的情况,收效特别明显,证明该方法的有效性。在青岛海试中恢复后信号的相关系数平均值为0.933,在青岛海试中利用这种解卷积技术去恢复信号是成功的。

When environment is complicated, original signals may be distorted and spread in time after propagation from source to receive array. Reconstructing an unknown signal from remote measurements made in an unknown and potentially complicated wave-propagation environment is needed in many applications. A signal recovery method for blind deconvolution is formally described based on a blind deconvolution technique in this paper. For a vertical array in an ocean sound channel, Green＇s function can be substituted with an artificial one, and a nonlinear operation can be changed to a linear operation, as a result of which, formulation of recovery or reconstruction of undistorted broadband signals from recordings made in unknown complex multipath environments can be deduced. The results of acoustic field numerical calculation for 6 types of sediment show that the correlation of a reconstructed broadband signal with the original signal is greater than 0.945 for each type, and the recovery performance is high especially for hard sediment. A Qingdao ocean experiment is carried out to investigate the performance of the signal recovery method for blind deconvolution of a broadband point-source signal from the propagation characteristics of an unknown oceanic waveguide. The results show that the method is successful and promising; the correlation between thereconstructed signal and the original signal is more than 0.933, which demonstrates that the recovered signal is undistorted.