基于3D融合特征联合神经网络的水声目标识别  

Hydroacoustic Target Recognition Based on 3D Fusion Feature Joint Neural Network

作　　者：许玮婷 赵英亮[1] 冯思奇 韩星程 贾彩琴 XU Wei-Ting;ZHAO Ying-Liang;FENG Si-Qi;HAN Xing-Cheng;JIA Cai-Qin(School of Information and Communication Engineering,North University of China,Taiyuan 030051,China;Jinzhong Branch,China Mobile Communications Group Shanxi Co.Ltd.,Jinzhong 030600,China)

机构地区：[1]中北大学信息与通信工程学院,太原030051 [2]中国移动通信集团山西有限公司晋中分公司,晋中030600

出　　处：《计算机系统应用》2025年第3期72-84,共13页Computer Systems & Applications

基　　金：国家自然科学基金青年科学基金(62203405);山西省应用基础研究计划(20210302124545,202303021212206,202202110401015)。

摘　　要：面对复杂的海洋环境,利用舰船辐射噪声进行水声目标特征提取与识别具有极大的挑战性.本文首先将船舶音频信号的三维梅尔频率倒谱系数(3D dynamic Mel-frequency cepstrum coefficient,3D-MFCC)特征与三维梅尔谱(3D dynamic Mel-spectrogram,3D-Mel)特征进行融合作为模型输入,并基于此提出了一种新的水声目标识别深度神经网络模型,该模型在卷积神经网络(convolutional neural network,CNN)和长短期记忆网络(long short-term memory,LSTM)的串行架构基础上,用多尺度深度可分离卷积网络(multi-scale depthwise convolutional network,MSDC),替代了传统的CNN,并增加了多尺度通道注意力机制(multi-scale channel attention,MSCA).实验结果表明,该方法在DeepShip数据集和ShipsEar数据集上的平均识别率分别达到了85.92%和97.32%,展现了良好的分类效果.Facing the complex marine environment,it is extremely challenging to utilize ship radiated noise for hydroacoustic target feature extraction and recognition.In this study,3D dynamic Mel-frequency cepstrum coefficient(3D-MFCC)features of ship audio signals are fused with 3D dynamic Mel-spectrogram(3D-Mel)features as model inputs.Based on this,a new deep neural network model for hydroacoustic target recognition is proposed.The model is based on the serial architectures of convolutional neural network(CNN)and long short-term memory(LSTM).Here,the traditional CNN is replaced by multi-scale depthwise convolutional network(MSDC),and multi-scale channel attention(MSCA)is added.The experimental results show that the average recognition rate of this method on DeepShip and ShipsEar datasets reaches 85.92%and 97.32%respectively,which demonstrates a good classification effect.

关 键 词：舰船辐射噪声 3D特征融合 多尺度深度可分离卷积 多尺度通道注意力机制 

分 类 号：TB566[交通运输工程—水声工程] TP183[理学—物理] TN929.3[理学—声学]