摘要
针对“黑飞”无人机飞行轨迹低、体积小、运动速度慢,其产生的信号回波数据量有限,且时域信号分辨率极低,传统雷达探测设备难以捕获“黑飞”无人机的特征反馈信息,论文提出了一种基于LSTM-YOLOv5的“黑飞”无人机异常行为辨识方法,利用YOLOv5算法实现无人机目标的跟踪识别,结合长短期记忆网络(Long short-term memory,LSTM)处理时间序列的优势,分析在该时间切片内的飞行趋势变化,以预测辨识“黑飞”无人机的异常行为。结果表明,论文提出的方法在无人机异常行为识别和分类方面具有较高准确性和鲁棒性,可有效应对“黑飞”无人机造成的威胁。
In view of the problem that the"black fly"drone has a low flying trajectory,small size,slow speed,and produces extremely low signal echo data volume and time domain signal resolution,and the feature feedback information is difficult to be cap⁃tured by traditional radar detection equipment,this paper proposes a method for identifying abnormal behaviors of"black fly"drones based on LSTM-YOLOv5.The YOLOv5 algorithm is used to track and identify the drone targets,and the advantages of LSTM in processing time series are combined to analyze the flight trend changes in the time slice to predict and identify the abnor⁃mal behavior of"black fly"drones.The results show that the proposed method has high accuracy and robustness in the identification and classification of abnormal behaviors of drones,and can effectively deal with the threats caused by"black fly"drones.
作者
袁江
兰增武
熊鹏
YUANG Jiang;LAN Zengwu;XIONG Peng(China Yangtze Power Company Limited,Yichang 443000)
出处
《舰船电子工程》
2023年第10期120-125,共6页
Ship Electronic Engineering
