论文部分内容阅读
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 estimation performance of Guaiyuan criteria source estimation algorithm in the case of small snapshots, a fuzzy clustering based estimation algorithm (FCBE) is proposed. The source number estimation algorithm based on fuzzy clustering uses the membership function value of the object to be classified (eigenvector) belonging to the signal subspace as the feature of the classification, and actually introduces a mapping from the object parameter to the classification feature, Deal with the actual situation of the estimated number of CITIC sources. The estimation performance of source number estimation algorithm based on fuzzy clustering and Gai’s circle criterion under small snapshot are compared by simulation experiments, and the practical application in shallow-water high-resolution sounding side-scan sonar verifies the performance of fuzzy clustering based on fuzzy poly The validity of the algorithm for estimating the number of sources of a class.