指数阻尼正弦模型阶选择算法

An Order-Selection Algorithm for Exponentially Damped Sinusoids Model

本文针对指数阻尼正弦信号模型提出了一种新的基于子空间和结构总体最小范数的阶选择算法.该方法利用信号子空间移不变性以及阶与信号分量数匹配时总体误差最小原理建立模型阶估计准则.实验表明该方法获得了更高的阶估计精度.算法无需设置阈值和惩罚项,自动完成阶估计过程.它可以应用到高分辨率的谐波恢复算法中.

This paper presents a new order-selection algorithm for exponentially damped sinusoids model based on subspace and structured total least norm method.It constructs model order estimation criterion by utilizing the shift invariance of the signal subspace,and the principle of total error reaching minimum when order matching with the number of signal components.The results of simulation show a higher estimation accuracy of the new algorithm.The algorithm is not required to set a threshold and penalization terms,and can accomplish estimation procedure automatically.It can be applied to the high resolution harmonic retrieval algorithms.