摘要
针对全面禁核试低频声监测中需要对大气低频声信号进行有效识别的问题,对深度神经网络中的卷积神经网络进行了研究,提出了一种将低频声信号转换为图像,然后采用卷积神经网络进行识别,并对学习过程进行改进的方法。将该方法与基于人工设计特征的支持向量机方法进行了对比实验,实验结果表明,在训练数据集不大的情况下,通过改进学习过程的卷积神经网络可以挖掘出信号的潜在特征,具有和支持向量机同等的识别能力。
To solve the problem of effective recognition of atmospheric low-frequency acoustic signal in the low-frequency acoustic monitoring of the total nuclear test,a method of using convolution neural network is proposed.It converts low-frequency acoustic signal into images,then puts images into convolution neural network.The method is compared with SVM method based on artificial design features.The experimental results show that,when the training data set is not large,the convolution neural network with improved learning process can mine the potential features of signals,it has the same recognition ability as SVM.
作者
吴涢晖
赵子天
陈晓雷
邹士亚
WU Yun-hui;ZHAO Zi-tian;CHEN Xiao-lei;ZOU Shi-ya(Research Institute of Chemical Defense,Academy of Military Sciences PLA,Changping Beijing,102205;School of Computer Science and Engineering,University of Electronic Science and Technology of China,Chengdu,611731)
出处
《电子科技大学学报》
EI
CAS
CSCD
北大核心
2020年第5期758-765,共8页
Journal of University of Electronic Science and Technology of China
关键词
大气低频声
卷积神经网络
深度学习
信号识别
支持向量机
atmospheric low-frequency sound
convolution neural network
deep learning
signal recognition
support vector machine
