多部雷达信号的复合分层异常挖掘研究  

A Study on the Outlier Mining from Multiple Sets of Radar Signals by a Composed Multi-level Method

作　　者：王文松[1] 李竹渝[2] 朱成文[2] 鲁万波[3] 

机构地区：[1]信息综合控制国家重点实验室,四川成都610036 [2]四川大学数学学院,四川成都610064 [3]西南财经大学统计学院,四川成都610074

出　　处：《无线电工程》2008年第9期13-15,共3页Radio Engineering

基　　金：信息综合控制国家重点实验室项目"海量数据信息挖掘与信息融合技术"(9140C1002040701)

摘　　要：对截获的雷达信号进行准确的分析和处理,实现对特定目标的异常特征信息进行快速发现、分析和挖掘,在提高雷达信号数据质量、发现入侵行为、进行故障诊断以及辅助军事侦察等领域都具有非常重要的意义。结合雷达信号的异常流行为,利用距离与聚类的统计思想,提出了一种针对雷达信号5个电子脉冲参数的复合分层分析异常挖掘方法,并运用5部仿真雷达的模拟数据进行了验证,发现该方法可对信度较高。Accurate analysis and processing of intercepted radar signals, and fast discovery, analysis and mining of unusual signs are very important to raising the quality of signals, finding the invasions and diagnosing the failure. Based on the five electronic features of the radar signals, this paper proposes a composed multi-level method to mine the outlier signals, combined with the unusual radar-signal flow, using the distance and clustering statistical thoughts. The verification is made by the simulation for 5 radar signals. The results show the method has higher reliability.

关 键 词：异常信号 数据挖掘 复合分层分析 脉冲参数 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]