雷达属性散射中心模型的属性参数能够提供目标更为丰富的重要信息,属性散射中心参数估计对解析雷达目标有着极其重要的研究意义.针对雷达属性散射中心模型,提出了基于深度学习的雷达属性散射中心快速目标分类和参数估计的技术.首先利用ViT(vision transformer)深度学习网络将雷达属性散射中心分类为局部式和分布式两类,然后基于TS2Vec框架构建针对属性散射中心参数估计的卷积神经网络(convolutional neural network for attribute scattering centers,ASC-NN),最后分别对两种数据进行训练以实现局部式和分布式属性散射中心的参数估计.基于属性散射中心模型展开数值实验,实验结果表明,该方法对雷达属性散射中心目标分类的准确率高达99%以上;雷达属性散射中心参数估计的速度超过传统方法的10 000倍以上,且精度更高,验证了所提方法的有效性和优越性.
Fast target classification and parameter estimation of radar attribute scattering centers
The attribute parameters of the radar attribute scattering center model can provide richer and more important information about the target,and the attribute scattering center parameter estimation is of great research significance for resolving radar targets.Aiming at the radar attribute scattering center model,this paper proposes the technique of fast target classification and parameter estimation of radar attribute scattering center based on deep learning.Firstly,the vision transformer(ViT)deep learning network is used to classify the radar attribute scattering centers into two categories:Localized and distributed,Then a convolution neural network for parameter estimation of attribute scattering centers(ASCNN)is constructed based on TS2Vec framework,and finally the two kinds of data are trained separately for parameter estimation of localized and distributed attribute scattering centers.Based on the attribute scattering center model,numerical experiments are carried out,the experimental results show that the accuracy of this method for radar attribute scattering center target classification is over 99%.The speed of radar attribute scattering center parameter estimation is more than 10 000 times higher than that of traditional methods,and the accuracy is higher,which verifies the effectiveness and superiority of the proposed method.
NETL
NSTL
万方数据
