中文摘要：

为利用奇异协方差矩阵实现高分辨率谱估计，提出一种动态约束的非对角加载类的谱估计方法．该方法一方面根据奇异采样协方差矩阵的结构，在其值空间对导向矢量投影的误差进行动态约束，获得具有较低噪声增益的窄带滤波器,另一方面，该方法利用快拍问相关性，形成Capon加权矢量,将特定频率点处确定的窄带滤波器与相应的Capon加权矢量共同作用于数据矩阵，获得该兴趣频率点幅度估计,仿真及实验结果表明：在采样协方差矩阵奇异的情况下，该谱估计方法获得的谱估计在精度及分辨率上均明显优于其它基于奇异协方差矩阵的谱估计方法，谱估计结果能够较好地保持原谱结构，该方法是一种利用奇异采样协方差矩阵进行高分辨率谱估计的有效方法。

英文摘要：

For estimating spectrum with high resolution based on the singular sample covariance matrix,a spectral estimator without diagonally loading is proposed with dynamic constraint. On the one hand, according to the structure of the singular sample covariance matrix, the proposed method adjusts the constraint of the error of the steering vector projection on its range subspace, and then the narrow filter with low noise gain is obtained. On the other hand, the method utilizes the correlation among the snapshots to form the Capon weight. The designed narrow filter, which is centered on the frequency of interest, and the corresponding Capon Weight are applied on the data matrix to obtain the amplitude estimate at the frequency. Simulation and experimental results show that：in case of sample covariance matrix singular, the proposed spectral estimator has better estimation accuracy and resolution than the other estimators, which employ the singular sample covariance matrix for estimating spectrum, the estimated spectrum also efficiently maintains the mw structure of spectrum. Therefore, the proposed estimator is an efficient spectral estimation method with utilizing the singular covariance matrix.