基于粗糙集与RBF神经网络的雷达辐射源信号识别研究  

Study on Radar Emitter Recognition Based on Rough Sets and RBF Neural Network

作　　者：张政超[1,2] 关欣[1,3] 李应升[2] 

机构地区：[1]海军航空工程学院信息融合技术研究所 [2]中国人民解放军63891部队 [3]国防科学技术大学电子科学与工程学院

出　　处：《航空计算技术》2009年第4期18-21,26,共5页Aeronautical Computing Technique

基　　金：国家自然科学基金资助项目(60572161);国家自然科学基金资助项目(60672140);全国优秀博士论文作者专项资金资助项目(200443);教育部新世纪优秀人才支持计划(NCET-05-0912)

摘　　要：针对新体制、新用途雷达辐射源信号难以识别的特点,用粗糙集理论对雷达辐射源信号进行离散化、属性约简、规则提取,达到分类的目的。用粗糙K-均值聚类方法计算径向基神经网络(RBFNN)的聚类中心,然后用粗糙集理论约简得到的规则构建径向基神经网络对未知雷达辐射源信号进行识别。仿真结果表明,这种基于粗糙集与RBF神经网络的识别模型减少了识别冗余特征、简化了神经网络结构,能有效地识别雷达辐射源信号。With the development of new type and use radar, it is more difficult to recognize radar emitter signal. The radar emitter signal information is converted into discrete value in this paper. The attribute of radar emitter signal is reduced and the decision rules are extracted based on Rough Sets. Then the cluster center of RBF neural network is gain by Rough K- means cluster method. The RBF neural network is constructed with the help of decision rules extracted from information table. The simulation result shows this radar emitter recognition model base on Rough Sets and RBF neural network can cut down the redundant attribute, lessen the neural network structure and recognize radar emitter signal effec-tively.

关 键 词：雷达辐射源识别 粗糙集 粗糙K-均值 RBF神经网络 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TN953[自动化与计算机技术—控制科学与工程]