基于深度学习的雷达辐射源型号识别方法

Radar Emitter Type Identification Algorithm Based on Deep Learning

针对传统雷达辐射源型号识别依赖专家经验构建识别模型,识别特征粗放、不完备,难以准确识别复杂体制雷达的问题,提出了一种基于深度学习的雷达辐射源型号识别方法。首先对时域波形数据降维、对齐、采样等预处理;然后采用受限玻尔兹曼机(RBM)和深度置信网络(DBN)模型训练样本;最后分别采用K最近邻(KNN)、随机森林(RF)、支持向量机(SVM)三种分类器完成识别任务。实验采用9类雷达辐射源型号的外场数据验证算法的有效性,实验表明基于深度学习的雷达辐射源型号识别方法取得了较好的识别效果。

Aiming at the problem that traditional radar emitter type identification relies on expert experience to construct identification model,and the identification feature are extensive and incomplete,so it is difficult to identify complex radar system accurately,a radar emitter type identification method based on deep learning is proposed.Firstly,the time-domain waveform data are pre-processed such as dimensionality reduction,alignment and sampling,then Restricted Boltzmann Machine(RBM)and Deep Belief Network(DBN)models are used to train the samples,and finally KNN,RF and SVM classifier are used to complete the identification task.The validity of the algorithm is verified by the outfield data of nine types of radar emitter.The experiment shows that a better identification effect could be achieved by the method of radar emitter type identification based on deep learning.