一种频率源噪声频谱拟合方法的研究

在分析频率源的噪声特性的时候,经常需要由Allan方差或功率谱计算频率源的五种噪声强度系数的值,通常采用人工分析的方法,求出噪声系数,这严重阻碍了自动化处理的应用。给出的一种自动拟合算法可以满足这方面的需要。在最小均方误差意义下,这种算法能自动完成对噪声类别的判定、噪声区间的划分以及噪声系数确定,所有这些无须人工参予。最后,对模拟数据和实测的几种频率源的功率谱曲线进行拟合,实验研究的结果表明这种方法拟合的相对误差不超过11％。

A Study of Fitting Method for Frequency Source Noise Power Spectrum

To analysis the noise character of frequency source, one must calculate the intensity coefficient of noise via Allan variance or power spectrum. The coefficient was usually calculated by manual work and this was a heavy obstruct in the automatic data processing. In this paper, an automatic fitting method is given. In the sense of least square error, this algorithm can determine the noise type,divide the range and estimate the intensity coefficient for every type of noise. At last,this method is checked with the simulated data and the measured data of several frequency sources, and the results show that the maximum relative error for this method is less than 11%.