长自回归谱的快速滤波算法

本文提出一种快速递推滤波算法,用于估计自回归(AR)模型系数。数值试验表明高分辨率的功率谱都是由高阶AR模型构成,高阶范围内有一最佳阶数存在,由于本算法是一种最优滤波算法,对噪声不敏感,小数据量时优于Marple算法,大数据量时优于快速付利叶变换,从文献记载的结果以及我们计算的实例看,有较高的分辨率。从本算法原理还能派生出若干类似算法。

A Fast filtering Algorithm for Higher Order Autoregressive Model

This paper gives a fast recursive filtering algorithm for Autoregressive model. numerical tests demonstrate that higher resolution power spectrum be consists of higher order autoregressive model, and There is the best order Which need to be found. the algorithm is both resist of noise , and is better than Marple algorithm in small number dataes and fast furier transformation in large number dataes, we can deduce many other similar algorithms from the algorithm.