多波束融合BP神经网络拖船噪声抵消  

Cancellation of noise from towing ship using BP neural network

作　　者：刘本奇[1] 杜选民[1] 

机构地区：[1]中船重工集团第七二六研究所,上海201108

出　　处：《声学技术》2007年第5期928-932,共5页Technical Acoustics

摘　　要：拖曳线列阵声纳中,拖船噪声常常是影响声纳性能的重要因素之一。由于存在严重的多途效应,仅仅改变自适应结构难以提高拖船噪声抵消性能。因此在实际应用中,固定参考输入自适应干扰抵消技术的抵消效果一般均不理想,本文针对拖船噪声的特性,提出了一种多干扰波束融合的抵消算法,该算法具有抗多途效应效果好,可以较有效地消除拖船干扰,提高声纳的检测性能,试验数据处理结果表明它优于现有的自适应方法,且具有运算量少的优点,可望在实际声纳中得到应用。In towed line array sonar, noise from the towing ship is an important factor often limiting the sonar performance. Multipath of the towing ship noise limits the performance of adaptive interference canceling （AIC） which utilizes the fixed beam as a reference input, This paper provides a new algorithm based on BP networks, which uses multi-beam fusion as its input. The scheme can cancel multipath towing ship noise with a low computational complexity. Results based on a large amount of trial data show that it is better than AIC.

关 键 词：拖船噪声 波束形成 干扰抵消 BP神经网络 

分 类 号：TN911.7[电子电信—通信与信息系统]