一种新的平稳随机信号功率谱估计法

A New Method of Parameter Estimation of Random Signals with ARMA Models

本文在随机过程ARMA模型参数的估计上提出了一种新方法.新算法基于超定MYW方程求解AR系数,修正了J.A.Cadzow提出的目标函数和加权函数;并推出一种分级白化估谱法以获得初始滤波参数,这是一种新的谱估计概念,结构简单.新算法的计算量远小于Cadzow法,实验结果显示了优良的谱估计性能.

This paper has presented a new method of parameter estimation of random signals with ARMA models. We have impreved the general objective and weighting functions to achieve good qualities in finding AR coefficients. To get the initial AR values, a novel concept of whitening in series is suggested. Experiments have shown that much higher resolutions are attained in estimation by our method, compared with cadzow's and other algorithms.