摘要
针对在复杂电磁环境下无人机难以被检测的问题,提出了一种基于卷积神经网络的无人机射频信号识别方法。射频前端对目标空域内的无线电信号进行扫描,捕捉与拦截无人机自身发射的射频信号,将无人机射频信号进行预处理,送入构建的卷积神经网络进行分析与识别。实验结果表明,基于卷积神经网络的无人机射频信号识别方法在检测无人机是否存在、识别4种无人机型号、识别10种无人机运行模式上均有较好的检测效果,具有较强的鲁棒性和环境抗干扰能力。
In complex electromagnetic environment,it is difficult to detect unmanned aerial vehicle(UAV).To solve the problem,a method of UAV radio-frequency(RF)signal recognition based on convolutional neural network is proposed.The RF front-end scans the radio signals in the airspace,and then the RF signal from the UAV itself is captured and intercepted.Finally,the RF signals of the UAV are preprocessed and sent to the constructed convolutional neural network for analysis and identification.Experimental results indicate the good performance of the method in detecting the existence of UAV and identifying four types and ten operating modes of UAV,and it is of strong robustness and environmental anti-interference capability.
作者
杨小伟
文清丰
杨雪
杨鹤猛
金熙
王泽跃
YANG Xiaowei;WEN Qingfeng;YANG Xue;YANG Hemeng;JIN Xi;WANG Zeyue(City West Power Supply Branch,State Grid Tianjin Electric Power Company,Tianjin 300301,China;Electric Power Research Institute State Grid Tianjin Electric Power Corporation,Tianjin 300301,China;Tianjin Zhongwei Aerospace Data System Technology Co.,Ltd.,Tianjin 300301,China)
出处
《无线电工程》
北大核心
2022年第3期456-462,共7页
Radio Engineering
基金
天津市电力公司科技项目(KJ21-1-7)。
关键词
无人机检测
射频信号识别
卷积神经网络
UAV detection
RF signal recognition
convolutional neural network
