导航自定位辐射源衰落信号智能识别算法  

Intelligent recognition algorithm for fading signal of navigation self positioning emitter

作　　者：吉丰 胡江湖 Ji Feng;Hu Jianghu(Array and Information Processing Laboratory,College of Computer and Information,Hohai University,Nanjing 211100,China)

机构地区：[1]河海大学计算机与信息学院阵列与信息处理实验室,南京211100

出　　处：《国外电子测量技术》2020年第7期52-57,共6页Foreign Electronic Measurement Technology

摘　　要：针对辐射源信号传播中受到多径影响产生不同程度的衰落从而影响导航定位这一问题,设计了一种基于频谱特征图的卷积神经网络(CNN)的信号衰落程度识别算法,并将该算法应用到导航自定位中来提升定位精度。该算法采用CNN提取衰落信号频谱特征进行训练和实现分类,并将采用算法选取和随机选取的辐射源信号分别应用到自定位场景中,对比分析所提算法的定位精度。结果表明,基于频谱特征图的CNN的衰落识别算法具有较好的分类性能,识别率高达90%以上,同时采用该算法使得测向标准误差下降了约0.03°,有效地提高导航自定位精度。Aiming at the problem that the radiation source signal is affected by multipath to produce different degrees of fading,which affects the navigation and positioning,this paper designs a signal fading degree recognition algorithm based on the convolutional neural network(CNN)of spectrum characteristic graph,and applies the algorithm to the navigation and self positioning to improve the positioning accuracy.In this algorithm,CNN is used to extract the spectrum features of fading signals for training and classification,and the emitter signals selected by algorithm and randomly selected are applied to the self positioning scene respectively,and the positioning accuracy of the recognition algorithm is compared and analyzed.The results show that the CNN fading recognition algorithm based on spectrum feature map has good classification performance,and the recognition rate is over 90%.Meanwhile,the algorithm reduces the direction finding error by about 0.03°,which effectively improves the navigation self positioning accuracy.

关 键 词：导航自定位 衰落信号识别 卷积神经网络 频谱特征图 

分 类 号：TN961[电子电信—信号与信息处理]