Partial EIV模型的非负最小二乘方差分量估计

Non-negative Least Squares Variance Component Estimation of Partial EIV Model

Partial Errors-in-Variables(Partial EIV)模型是EIV模型的扩展形式,权阵构造简单,当系数矩阵中存在非随机元素和随机元素时,Partial EIV模型的适用性更强。针对Partial EIV模型中随机模型不准确的情况,将系数矩阵和观测向量分别作为一类数据,本文在该模型的基础上,使用最小二乘方差分量估计方法,推导相关计算公式及迭代算法,分别估计出相应的方差分量估值。并对出现的负方差使用非负最小二乘理论,增加约束条件,对随机模型进行修正,得到更加合理的参数估值。试实验结果表明,本文的方法与其他方差分量估计方法等价。

As an extended form of the errors-in-variables(EIV)model,and the weight matrix is easy to structure,the applicability is stronger when both non-random elements and random elements exist in the coefficient matrix.According to the inaccurate stochastic model in the Partial EIV model,the coefficient matrix and observation vector are used as a kind of data respectively.The least squares variance component estimation method based on Partial EIV model is used and the relevant calculated formulas and iterative algorithm are derived.Then the corresponded variance components are estimated.The non-negative least squares is used when the negative variance appears,then add constraint condition to correct the rand model,so the estimated parameters are more reasonable.The experiments show that the results obtained by the method of this paper and other methods are equivalent.