奇异值分解在海洋回波多普勒信息提取中的应用

运用矩阵奇异值分解原理，分析了功率谱估计对探测海洋表面信息和硬目标信号的意义，对接收信号的自相关矩阵进行分解以提高谱估计的分辨率，并将处理后的方法对实测数据进行功率谱的估计，与处理前的回波谱图进行比较，实验证明，得出的奇异值分解方法可以改善多普勒回波功率谱的性能，满足高频地波雷达的目标探测要求。

Application of SVD in the Estimation of Sea Echo Doppler Power Spectra

From the sea echo Doppler power spctra we can get not only state parameter of surface current but also some useful information of the moving target on the surface of the sea.In this paper,the application ofpower spectrum estimation in the detection of target and state of surface current is investigated.Because of the inevasible contradiction between the resolution and stability of spectrum,almost all traditional methods have an inherenet defect that they could not provide high enough resolution to pick up ship echo from sea clutter.However, most modern estimation methods are sensitive to the accurcy of signal covariance matrix,so the whole method can be improved by transforming the signal covariance mtrix.On demand of the signal detection, Sigular Value Decomposition theory is used in the estimation of sea echo Doppler power spectra. According to the Sigular Value Decomposition theory, the recieved signal by the radar can be divided into two subspaces: the signal subspace and the noise subspace. The two subspaces are orthogonal, so the product of eigenvectors associated with the smallest eigenvalues and signal are zero. That is to say the frequency of signal is correspondnt to the peak in the power spectrum.In the case of three modern estimation methods, either signal subspace or noise subspace is reserved, which can lead to the improvement of performance. By using the data of the experiment in Oct. 2000,the result of the improved methods is compared with that of the original ones in the aspects of performane and stability.The conclusion that SVD theory can improve the performance of estimation of sea echo Doppler spectrun is drawn. And we can expect that such improvment can satisfy the requirment of movig targets detection of HF ground wave radar.