噪声背景下正弦信号频率估计:基于约束广义逆的总体最小二乘方法

Estimation of Sinusoidal Frequencies in Noise: SVD-TLS Based on Constrained Generalized Inverse Matrix

分析了噪声背景下实谐波过程ARMA模型系数之间的对称性,并以此为约束条件加入到ARMA谱估计方法的求解过程中,从而提出了一种改进的正弦信号频率估计方法。理论分析与计算机仿真表明,对于低信噪比条件下的正弦信号参量估计,这一算法的分辨率与精度都优于MUSIC方法和仅使用总体最小二乘法(SVD-TLS)的ARMA谱估计方法。

In this paper symmetry among coefficients of ARMA model of real harmonic process in noise is analyzed and added into the estimation process of sinusoidal frequencies based on ARMA method as constraint condition. Theoretical analyses and computer simulation shows this method performs better in resolution and precision of estimation of sinusoidal frequencies in strong background noise than MUSIC method and ARMA method that use only SVD-TLS.