一种新的未知雷达信号聚类分选方法

摘要

The k-means clustering algorithm is considered that it is not applicable to unknown radar signal sorting.A sorting method based on improved grid clustering is proposed.Signal grid and noise grid are distinguished using the double-density threshold,after data points are mappinged to grid space.Finally,the signal sorting is achieved through the high density connectivity of grids.In addition,the signal sorting without any prior knowledge could be accomplished,and this method is unsensitive to the noise.The simulation results validate the effectiveness of the method.%通过分析,说明了k-means聚类算法不适用于未知雷达信号分选,进而提出一种改进网格聚类分选方法。该方法将数据点映射到网格空间,通过双密度阈值准确区分信号网格与噪声网格,利用网格的高密度连通性完成聚类分选。该方法计算量小,对噪声不敏感,无需先验知识支持。计算机仿真验证了算法的有效性。