训练样本不足时雷达扩展目标检测方法  

Radar range-spread target detection with limited training data

作　　者：黎炎 李哲 陈扬[1,2,3] 王剑 胡丹晖 吴驰 Li Yan;Li Zhe;Chen Yang;Wang Jian;Hu Danhui;Wu Chi(NARI Group Corporation,Nanjing 211106,China;Wuhan NARI Limited Liability Company of State Grid Electric Power Research Institute,Wuhan 430074,China;Hubei Key Laboratory of Power Grid Lightning Risk Prevention,Wuhan 430074,China;State Grid Corporation of China,Beijing 100031,China;State Grid Hubei Electric Power Company,Wuhan 430074,China;State Grid Sichuan Electric Power Company,Chengdu 610041,China)

机构地区：[1]南瑞集团有限公司,江苏南京211106 [2]国网电力科学研究院武汉南瑞有限责任公司,湖北武汉430074 [3]电网雷击风险预防湖北省重点实验室,湖北武汉430074 [4]国家电网有限公司,北京100031 [5]国网湖北省电力公司,湖北武汉430074 [6]国网四川省电力公司,四川成都610041

出　　处：《南京理工大学学报》2018年第6期727-731,共5页Journal of Nanjing University of Science and Technology

基　　金：国家电网公司科技项目(524625160016)

摘　　要：为了降低对训练样本的需求,针对雷达扩展目标检测问题,该文提出了降秩广义似然比检验(R-GLRT)检测器和降秩Wald(R-Wald)检测器。利用噪声子空间对应的特征矩阵代替采样协方差矩阵,降低了训练样本不足时小训练样本带来的估计误差。仿真结果表明,当训练样本不足时,所提出的降秩检测器能够提供较高的检测概率,且R-GLRT检测器具有比R-Wald检测器更高的检测概率;当训练样本充足时,与常规自适应检测器相比,2种降秩检测器也能够提供较高的检测概率。A reduced-rank generalized likelihood ratio test(R-GLRT)detector and a reduced-rank Wald(R-Wald)detector are proposed for radar range-spread target detection to reduce the requirement of training samples.The sampling covariance matrix is replaced by a characteristic matrix corresponding to the noise subspace,and the estimating error is reduced for low sample support.Simulation results show that the proposed reduced-rank detectors can work properly with low sample support,and the detection performance of the R-GLRT detector is better than that of the R-Wald detector;the detection performance of the proposed reduced-rank detectors is better than that of conventional adaptive detector with sufficient training data.

关 键 词：雷达 扩展目标 信号检测 降秩 广义似然比检验 

分 类 号：TN957[电子电信—信号与信息处理]