摘要
最大熵谱估计法是一种高分辨率的现代谱估计法。传统谱估计法的分辨率最好也不超过1/T,其中T为数据窗长度。因此当数据长度小于信号周期时就无法计算出信号的实际频率。而最大熵谱估计法取消了窗函数,从而也没有基频的概念。由于用随机过程的自相关序列进行无限外推可获得更加逼近的谱,因此大大提高了谱的分辨率。 本文进一步研究了最大熵谱估计中振幅失真的解决办法,即用一经验公式作为门槛值来选择最佳频率成分,以提高最大熵谱估计的使用能力,力求在油气检测及岩性分析中得以应用。
Abstract
The maximum entropy estimation is one of the modern high resolution spectrum estimation methods. The highest resolution of the conventional spectrum estimation is not beyond the limit of 1/T, where T is the length of the data window. For this reason, when the data length is less than that of the signal period, it is impossible to extract the actual frequency of the signal. But for the maximum entropy estimation, because of the window function is not used, hence, it has nothing to do with the concept of the base frequency. The maximum entropy estimation, by using the autocorrelation sequence of the random process to make the infinite extrapolation, therefore, a more approximate spectrum could be obtained and the resolution of the spectrum are hence improved.In this paper, the way for solving the problem of amplitude distortion in the maximum entropy estimation was examined. In frequency selection, by using an empirical formula as the threshould, the result of the maximum entropy estimation could be visibly improved. It could be expected that the technique above mentioned may find its way in actual oil/gas detection and lithological analysis.
论文详情