用多项式模型和自适应滤波估计瞬时频率

本文论述了非平稳信号瞬时频率(IF)的估计方法,包括基干最小均方误差(LMS)算法或递推最小二乘(RLS)算法的自适应IF估计方法和基于信号相位多项式模型的IF估计方法。对各种估计方法的估计方差与理论方差下限进行了比较。说明了,基于互Wigner分布的估计方法所需信噪比(SNR)门限值最低。在低SNR、短信号记录的情况下,用多项式相位模型与最大似然法可以获得比较可靠的IF估计。

Estimation of Instantaneous Freguency Using Polynomial Model and Adaptive Filtering

The methods for estimating the instantaneous frequency (IF) of a nonstationaty signal is dis- cussed. These methods include adaptive estimation methods based on the least mean squares (LMS) error algorithm, or the recursive Least squares (RLS) algorithm, and methods based on a modeling of the signal phase as a polynomial. A comparision of the various methods with one another and with the theoretical lower variance bounds has been made. It has been showed that the method based on a cross Wigner distribution has the lowest signal-to-noise ratio (SNR) threshold. For short signals with low SNR, a combination of polynomial phase modeling and maximum likelihood techniques can make reliable estimates.