SEResNet17 method suitable for dynamic environment underwater acoustic target recognition

The invention provides an SEResNet17 method suitable for dynamic environment underwater acoustic target recognition, which comprises the following steps: firstly, extracting underwater acoustic signal features by using a residual network, eliminating a gradient disappearance phenomenon in an optimiz...

Full description

Saved in:
Bibliographic Details
Main Authors XUE LINGZHI, ZENG XIANGYANG
Format Patent
LanguageChinese
English
Published 08.04.2022
Subjects
CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
HANDLING RECORD CARRIERS
PHYSICS
PRESENTATION OF DATA
RECOGNITION OF DATA
RECORD CARRIERS
Online AccessGet full text

Cover

More Information
Summary:The invention provides an SEResNet17 method suitable for dynamic environment underwater acoustic target recognition, which comprises the following steps: firstly, extracting underwater acoustic signal features by using a residual network, eliminating a gradient disappearance phenomenon in an optimization process by using the residual network, analyzing the output of the residual network, further finding out a channel where feature data effective for a recognition task is located, weighting the channel, and finally obtaining a target recognition result; and the accuracy of underwater sound recognition is improved. 本发明提出一种适用于动态环境水声目标识别的SE_ResNet_17方法，SE_ResNet_17模型首先利用残差网络提取水声信号特征，残差网络消除优化过程中的梯度消失现象，经分析残差网络的输出，进一步找到对识别任务有效的特征数据所在的通道，对通道进行加权，提高水声识别的正确率。
Bibliography:Application Number: CN202111512006