论文部分内容阅读
在地球物理学领域中,存在着大量谱估计问题。常用的方法有:间接法、相关变换法和付里叶变换等。这些方法总是给资料加上某种限制,使作出的谱估计分辨率低,谱值出现畸变,而极大熵谱法(Maximum Entropy Metbod,简称MEM)克服了这种缺点。熵是随机过程单位时间(或自变量单位间距)的平均信息,而平稳过程的熵取极大时的功率谱即为最大熵谱。它可以得到一种高阶AR(P)模型来描述时间序列。它对功率谱的估计,尤其对卓越周期的寻求比传统方法要好;能给出分辨率高,功率偏移小的估计。图1就是非常好的例证。
In the field of geophysics, there are a large number of spectral estimation problems. Common methods are: indirect method, the relevant transformation method and Fourier transform. These methods always impose some restrictions on the data to make low spectral resolution and spectral distortion, and the Maximum Entropy Metabod (MEM) overcomes this disadvantage. Entropy is the average information of random process unit time (or unit distance of independent variables), while the power spectrum when the entropy of the stationary process is maximal is the maximum entropy spectrum. It can get a high-order AR (P) model to describe the time series. Its estimation of the power spectrum, especially for excellent cycles, is better than the traditional method; it gives an estimate of high resolution and small power offset. Figure 1 is a very good example.