摘要
针对线性调频雷达系统中的大数据量采集对模数转换及后续数据存储造成较高负担的问题,提出了一种在欠采样条件下、基于卷积神经网络的信号重构方法。首先设置两个不同的位置和速度的运动目标,利用线性调频信号对目标进行探测,对接收端产生的差拍信号进行欠采样;其次构建重构网络模型并利用欠采样数据对该网络进行迭代训练;最后测试该网络对差拍信号的重建能力。结果表明,该方法在5倍欠采样的条件下,可重构出原始差拍信号并能以较低误差提取目标位置与速度信息,有效减小了系统所需采集量,对提高信号处理效率具有重要作用。
In order to solve the problem that large data acquisition in Linear frequency modulated radar system has a high burden on analog-to-digital conversion and subsequent data storage,a signal reconstruction method based on convolutional neural network under under-sampled condition is proposed.Firstly,two moving targets with different positions and velocities are set up,the target is detected by linear frequency modulation signal,and the differential beat signal generated by the receiver is under-sampled.Secondly,the reconstruction network model is constructed and the network is trained iteratively by under-sampling data.Finally,the reconstruction ability of the network to the differential beat signal is tested.The results show that the method can reconstruct the original differential beat signal and extract the target position and velocity information with lower error under the condition of 5 times under-sampling,which can effectively reduce the required acquisition amount of the system.It plays an important role in improving the efficiency of signal processing.
作者
余晨
杨振泽
谷建星
景宁
Yu Chen;Yang Zhenze;Gu Jianxing;Jing Ning(School of Information and Communication Engineering,North University of China,Taiyuan 030051,China;Shanxi Provincial Research Center for Opto-Electronic Information and Instrument Engineering Technology,Taiyuan 030051,China)
出处
《国外电子测量技术》
北大核心
2021年第9期143-148,共6页
Foreign Electronic Measurement Technology
基金
国家留学基金委(201908140065)
山西省高等院校科技创新计划(201701D31100002)项目资助。
关键词
欠采样
卷积神经网络
线性调频
差拍信号
under-sampling
convolutional neural network
linear frequency modulation
beat signal
