| 基于卷积神经网络技术的直升机旋翼谱识别方法 |
| |
| 引用本文： | 王志诚,徐卉,梁小溪,王珺,周起华,朱骏.基于卷积神经网络技术的直升机旋翼谱识别方法[J].上海航天,2018(6):103-111. |
| |
| 作者姓名： | 王志诚 徐卉 梁小溪 王珺 周起华 朱骏 |
| |
| 作者单位： | 上海交通大学 电子信息与电气工程学院,上海 200240;上海无线电设备研究所,上海 201109,上海无线电设备研究所,上海 201109,上海机电工程研究所,上海 201109,上海无线电设备研究所,上海 201109,上海无线电设备研究所,上海 201109,上海无线电设备研究所,上海 201109 |
| |
| 基金项目： | 军委科技委国防科技项目 |
| |
| 摘 要： | 针对直升机雷达回波在时频二维分布图上的分布特性,提出一种基于卷积神经网络技术的直升机旋翼谱识别方法,实现了直升机旋翼谱与其他类型目标的区分,从而有效识别直升机。当雷达工作在低重复频率(LPRF)时,直升机旋翼回波在时频分布图上呈现明显的扩展特征。通过仿真与实测数据分析,表明该特征与其他目标回波在频率维分布特性存在较大差异。利用卷积神经网络中的卷积核技术,将实测数据与卷积核矩阵进行二维卷积运算,依据输出矩阵实现了直升机旋翼谱的识别。利用实际采集的直升机回波数据进行验证,证明该方法是有效的,可应用于各种采用脉冲多普勒工作体制的雷达系统。
|
| 关 键 词： | 卷积神经网络 雷达 目标识别 直升机旋翼 |
| 收稿时间： | 2018/9/15 0:00:00 |
| 修稿时间： | 2018/11/2 0:00:00 |
Identification Method of Helicopter Rotor Spectrum Based on Convolution Neural Network Technology |
| |
| WANG Zhicheng,XU Hui,LIANG Xiaoxi,WANG Jun,ZHOU Qihua and ZHU Jun.Identification Method of Helicopter Rotor Spectrum Based on Convolution Neural Network Technology[J].Aerospace Shanghai,2018(6):103-111. |
| |
| Authors: | WANG Zhicheng XU Hui LIANG Xiaoxi WANG Jun ZHOU Qihua and ZHU Jun |
| |
| Affiliation: | School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China;Shanghai Radio Equipment Research Institute, Shanghai 201109, China,Shanghai Radio Equipment Research Institute, Shanghai 201109, China,Shanghai Electro-Mechanical Engineering Institute,Shanghai 201109, China,Shanghai Radio Equipment Research Institute, Shanghai 201109, China,Shanghai Radio Equipment Research Institute, Shanghai 201109, China and Shanghai Radio Equipment Research Institute, Shanghai 201109, China |
| |
| Abstract: | In view of distribution characteristics of the helicopter radar echo on time-frequency two-dimensional histogram, a helicopter rotor spectrum recognition method based on convolution neural network technology is proposed. The spectra can be distinguished from helicopter rotor to other types of targets, and the helicopter can be effectively identified. When the radar operates in low pulse repetition frequency (LPRF) condition, the helicopter rotor echo presents obvious extension characteristics in the time-frequency histogram. This characteristic is quite different from those of other target echoes on frequency dimensional distribution. The convolution kernel technique in convolution neural network is used to convolute the measured data and convolution kernel matrix in two dimensions, and to realize the helicopter rotor spectrum recognition based on the output matrix. The effectiveness of the method is verified by the actual helicopter echo data collected. The method can be applied to various radar systems with pulse Doppler operating system. |
| |
| Keywords: | convolution neural network radar target recognition helicopter rotor |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《上海航天》浏览原始摘要信息 |
|
点击此处可从《上海航天》下载全文 |
|
