基于自回归协方差阵的似乎不相关模型参数的线性贝叶斯估计

The Linear Bayesian Estimation of Parameters in Seemingly Unrelated Model Based on Covariance Autoregressive Matrix

摘要: 在本文中，构造了已知协方差矩阵结构的均值–协方差似乎不相关模型，该均值–协方差模型考虑了具有时间相关性的自回归模型，并得到该模型参数的线性Bayes估计。在均方误差矩阵准则下对与广义最小二乘估计方法和线性Bayes估计方法进行了对比，通过模拟验证了相对于广义最小二乘估计方法的优良性。

Abstract: In this paper, we propose a mean-covariance seemingly uncorrelated model with known covariance structures. The mean-covariance model is used to describe the autoregressive model with time correlation, and the parameters of this model were estimated by using linear Bayesian estimate. Under the criterion of mean square error matrix, the generalized least squares estimation method and the linear Bayes estimation method were compared; with respect to the generalized least squares estimation method, the superiority of Bayes Linear Estimator is verified by simulation.