一种严格最大熵谱估计法的递推算法

A Strict Maximum Entropy Spectral Estimation and Its Recursive Algorithm

本文提出了基于一种更严格最大熵谱估计的快速递推算法。讨论了当数据长度有限时，为满足最大熵谱估计的四个近似等价条件，所提出的严格目标函数。在此基础上，通过使目标函数达到极小，用递推法估计出各阶反射系数，然后由Levinson递推得到最大熵谱参数的估计值。该算法算法稳定，快速。计算机仿真表明，此估计方法具有良好的估计质量。

This paper presents a new recursive algorithm on a more strict maximum entropy spectral estima tion. The more stringent MESE can make all the four approximately equivalent hold by defining a new object function that expreses synthetically both the energies of forward and backward linear prediction errors and the orthognnality relation between the predictionn errors and the signal. Not estimating the MESE parameters directly, the MESE reflection cocfficients are estimated firstly, then by the light of the Levinson recursion,the MESE parameters can obtained quickly. The reflection coefficient estimates are obtained by minimizing the object function for different order in a new recursive manner,whose alorithm is steady.Simulations verify that the behavior of its estimates is better than the Burg method.