一类非高斯噪声干扰信号的广义TLS-ESPRIT谱估计算法

提出了一类比较普遍存在的非高斯噪声－过滤白噪声的定义，并且讨论了这类噪声的性质。在此基础上，本文拓宽了对信号噪声的要求，对传统的ＴＬＳ－ＥＳＰＲＩＴ谱估计算法进行一定的改进，推导出了在这类噪声加性干扰下信号的广义ＴＬＳ－ＥＳＰＲＩＴ谱估计方法。通过特征分解技术，获得并证明了一种有实际应用价值的算法。这种算法具有高分辨、运算量低、短数据和宽频带等特点。该算法对噪声的假设更加一般化，具有比其他方法更强

ESPRIT Method of Parameter of Sinuasoids in a Series of Non-Gauss Noise

A group of noises, the filtered white noise, is defined and their characters are studied, which is more general and non Gaussian. The request of the basic TLS ESPRIT method is aggrandized and its algorithm approved. The TLS ESPRIT for spectrum estimation is presented and the signal with the filtering white noise added. A new applied algorithm is obtained through the eigendecomposition technology. The new algorithm in this paper has high resolution, short dada record and wide band characteristics. It is more practical for the wide supposition than other methods. A practical example shows the validated performances of the method and illustrates the correctness and practicability of the algorithm.