基于加权SVC和K-Mediods联合聚类的雷达信号分选方法  

Radar Signal Sorting Method Based on Weighting SVC and K-Mediods Combined Clustering

作　　者：吴连慧[1] 秦长海[1] 宋新超[1] 

机构地区：[1]中国船舶重工集团公司第七二三研究所,江苏扬州225001

出　　处：《舰船电子对抗》2017年第1期13-17,共5页Shipboard Electronic Countermeasure

摘　　要：为了提高复杂体制雷达信号分选的正确率,提出了加权SVC和K-Mediods联合聚类算法,针对雷达参数特点,对SVC算法的核函数内积和K-Mediods算法的欧氏距离进行加权计算,从而避免聚类结果被弱相关的特征所支配。与SVC与K-Means联合聚类算法相比,SVC与K-Mediods联合聚类算法有效降低了"离群点"的影响。结果表明,该算法能够提高复杂体制雷达信号分选的正确率,存在部分"离群点"时分选正确率较高。To improve the accuracy of complex system radar signal sorting, this paper presents the algorithm of weighting SVC and K-Mediods combined clustering, performs the weighting calcula- tion to kernel function inner product of SVC algorithm and Euclidean distance of K-Mediods algo- rithm aiming at the characteristics of radar parameters, accordingly avoids the clustering result is dominated by the characteristic of weak correlation, compares SVC and K-Mediods combined clus- tering algorithm with SVC and K-Means combined clustering algorithm, and the former decreases the influence of outlier effectively. Results indicate that this algorithm can improve the accuracy of complex system radar signal sorting,the sorting accuracy is relatively higher when part outliers exist.

关 键 词：雷达信号分选 支持向量聚类 K-中心点 权值 

分 类 号：TN971.1[电子电信—信号与信息处理]