基于误差反向传播算法的多装侦察数据型号识别方法  

A Recognition Method for Multi-Reconnaissance Data Based on Error Back Propagation Algorithm

作　　者：张译方 旷生玉 梁璟 徐晶 ZHANG Yi-fang;KUANG Sheng-yu;LIANG Jing;XU Jing(Science and Technology on Electronic Information Control Laboratory,Chengdu 610036,China)

机构地区：[1]电子信息控制重点实验室

出　　处：《电子信息对抗技术》2019年第6期18-22,60,共6页Electronic Information Warfare Technology

摘　　要：为解决广域战场空间中,多装侦察数据、混叠参数的型号识别问题,提出一种基于误差反向传播算法的多装侦察数据型号识别方法。通过对各类装备信息进行建模,考虑不同装备的截获误差,进而能够自适应生成识别算法进行数据处理。进一步,提出基于误差反向传播算法的型号识别方法,解决参数混叠下的型号识别问题。工程实践表明,该方法能够规避不同装备的测量误差影响,普遍提升型号识别的准确度;同时,该方法通过样本数据不断训练网络模型,显著提升特征参数混叠情况下的识别率。To solve the radar recognition problem of multi-reconnaissance data and interleaving parameters in wide battlefield areas,a scheme based on Error Back Propagation(BP)which can recognize multi-reconnaissance data is proposed.Firstly,the reconnaissance data of various equipment is modeled in consideration of different equipment reconnaissance capture errors,so that recognition algorithms can be automatically generated while data processing.Secondly,a method based on BP is proposed to solve the radar recongnition problem with interleaving parameters.Engineering practice shows that a general gain of accuracy can be achieved for radar type recognition in avoidance of errors for different equipments.Meanwhile,the network parameter configurations can be continuously trained and optimized,to achieve an iterative optimization for radar recognition.

关 键 词：雷达型号识别 神经网络 误差反向传播 

分 类 号：TN971.1[电子电信—信号与信息处理]