基于加权SVC和K-Mediods联合聚类的雷达信号分选方法

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 calculation to kernel function inner product of SVC algorithm and Euclidean distance of K-Mediods algorithm 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 clustering 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.