摘要
水下安静型目标主动探测与识别问题中,确定目标与发射装置的相对姿态是目标探测的关键。本文结合水下目标声散射回波机理,对目标的声散射回波成分进行信号特性分析,结合分数阶傅里叶变换对声散射回波进行特征提取;利用循环神经网络保存目标回波时间序列信息,实现对序列信号的动态建模;分别将分数阶傅里叶域特征与频谱结构特征作为训练对象,构建水下目标回波角度识别模型并进行对比,实现对不同声波入射角度下目标回波的分类识别。水池实验数据处理结果表明:本文提出的分数阶傅里叶特征与长短期记忆系统结合的深度神经网络模型,各项评估指标均优于以频谱结构为训练特征的深度网络模型,验证了本文方法对目标姿态识别的有效性。
The key to target detection in the active detection and recognition of quiet underwater targets is determining the relative attitude of the target and the launcher.In this paper,the signal characteristics of a target acoustic scattering echo components are analyzed according to the target acoustic scattering echo mechanism.The characteristics of the acoustic scattering echo are extracted using the fractional Fourier transform method.A recurrent neural network is employed to save the time-series information of the target echoes,realizing dynamic modeling of sequence signals.The fractional Fourier domain features and spectral characteristics are trained to build and compare the underwater target echo angle recognition models.This way,the classification and recognition of the target echoes with different sound wave incident angles are realized.The data processing results of the water tank experiment in this study indicate that the proposed deep neural network model,which combines the fractional Fourier features,and the long and short-term memory system has advantages over that with spectral structure as the training object in all evaluation indicators,which verifies the effectiveness of the method for target pose recognition.
作者
李秀坤
徐天杨
嵇守聪
LI Xiukun;XU Tianyang;JI Shoucong(Acoustic Science and Technology Laboratory, Harbin Engineering University, Harbin 150001, China;Key Laboratory of Marine Information Acquisition and Security (Harbin Engineering University), Ministry of Industry and Information Technology, Harbin 150001, China;College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, China)
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2021年第10期1503-1509,共7页
Journal of Harbin Engineering University
基金
国家自然科学基金项目(11774073).
关键词
水下目标识别
目标回波
声波入射角度
特征提取
分数阶傅里叶变换
时序结构
深度学习
循环神经网络
underwater target recognition
target echo
incident angle of sound
feature extraction
fractional Fourier transform
temporal structure
deep learning
recurrent neural network
