基于模糊聚类信源数估计算法的研究、仿真与应用  

Research,Simulation and Application of Signal Number Estimator Based on Fuzzy Clustering

作　　者：胡隽[1] 

机构地区：[1]中国科学院声学研究所海洋声学技术实验室,北京100080

出　　处：《系统仿真学报》2008年第5期1133-1138,共6页Journal of System Simulation

基　　金：国家863计划资助项目(2005AA611020)

摘　　要：为了改善在小快拍情况下盖氏圆准则信源数估计算法的估计性能,提出了一种基于模糊聚类的信源数估计算法(Fuzzy Clustering Based Estimator-FCBE)。基于模糊聚类的信源数估计算法使用了待分类对象(特征向量)隶属于信号子空间的隶属函数值作为分类的特征,实际上引入了从对象参数到分类特征的一次映射,符合阵列信号处理中信源数估计的实际情况。通过仿真实验比较了基于模糊聚类的信源数估计算法和盖氏圆准则在小快拍下的估计性能,并且在浅水高分辨率测深侧扫声纳中的实际应用验证了基于模糊聚类的信源数估计算法的有效性。In order to improve the effectiveness of the Gerschgorin Disks Estimator when the number of snapshots is small, a Fuzzy Clustering Based Estimator-FCBE was suggested. FCBE exploited the membership of eigenvectors to the signal subspace as the character of clustering. Compared with ordinary Fuzzy clustering, FCBE included a mapping that was the key to apply the Fuzzy clustering method in signal number estimation. In computer simulation, the performance of FCBE with the performance of Gerschgorin Disks Estimator was compared. And in the signal processing of shallow water Bathymetric Side-scan Sonar System, FCBE could estimate the number of incoming signals efficiently and correctly.

关 键 词：模糊聚类 信源数估计 盖氏圆准则估计 波达方向估计 

分 类 号：TN911.23[电子电信—通信与信息系统]