一类非平稳噪声线性模型的估计及其在地壳形变中的应用

ESTIMATION OF A LINEAR MODEL WITH NON-STATIONARY NOISE AND ITS APPLICATION TO CRUSTAL DEFORMATION

本文将灰箱中的线性模型纳入随机模型来处理,充分考虑模型误差含有粗差且具有相关——自相关的情形,提出这一类非平稳噪声线性模型的误差模式。采用稳健估计法(Robust估计)和长自回归白噪化方法来获取模型的最优估计。同时以本文提出的误差模型及相应估计方法为基准,说明当前统计学界常见的误差模型及估计方法是其退化引出的。最后给出一实例,说明此方法的优越性。

In this paper,fully considering the case that the model error contains the gross error and the correlation and autocorrelation, the error model of a linear model with non-stationary noise is proposed by processing the linear model in the grey box brought into the random model. The optimal estimation of model is obtainod using the robust estimates and the high-order AR model methods.At the same time, based on the error model and the relative estimate method presented in the paper, it is shown that the existing error model and estimate ways on the statistic circles are drawn from their degeneration. Finally, a practical example is given to illustrate the advantage of this method.