基于AP密度聚类方法的雷达辐射源信号识别

未知雷达辐射源信号识别一直是雷达对抗情报分析中的难题。针对基于密度的聚类算法在处理不均匀样本时识别率较低的缺陷,将该算法与亲和传递（AP）聚类算法结合,提出一种基于AP密度聚类的识别方法。该方法先利用AP聚类方法对数据样本进行初步聚类,再设定相关参数,运用基于密度的带有噪声的空间聚类（DBSCAN）算法进行二次聚类。相对于原样本,初始聚类结果分布具有一定的代表性,容易找到适合DBSCAN方法的参数值。测试表明该方法具有较高的识别率。

Signal Identification of Radar Radiation Source Based on AP Density Clustering Method

Signal identification of unknown radar radiation source is always a problem of intelligence analysis of radar countermeasure.Aiming at the shortage that the identification probability is low when the clustering algorithm based on density is used to process non-uniformity samples,this paper combines the algorithm with affinity propagation(AP) clustering algorithm,brings forward an identification method based on AP density clustering method.The method firstly uses AP clustering method to perform the primary clustering to the data samples,then sets up the correlative parameters,uses the algorithm of density based spatial clustering of application with noise(DBSCAN) to perform secondary clustering.Comparing with original samples,the distribution of primary clustering results is representative and the parameter values adapted for DBSCAN algorithm can be found easily.The method is verified to have better identification probability through the test.