基于改进目标函数的partial EIV模型WTLS估计的新算法

A New Algorithm of Weighted Total Least Squares Estimate of Partial EIV Model Based on an Improved Objective Function

摘要: 针对部分变量误差（partial EIV）模型的加权整体最小二乘（weighted total least squares，WTLS）估值的计算需要多次迭代且效率低下的情况，根据加权LS（least square）原理，通过改进目标函数，并运用矩阵微分运算以及矩阵反演变换，提出了一种计算partial EIV模型WTLS估值的新算法。算例计算结果表明，新算法具有迭代次数少、计算效率高等优点。

Abstract: The computation of weighted total least squares (WTLS) estimate of partial EIV model requires more iterations and more computation burden. Therefore, the study has proposed a new algorithm for computing the WTLS estimate of partial EIV model by improving objective function based on the weighted LS principle and applying the differential and inversion transformation of matrix. The results of numerical examples show that the new algorithm requires less iterations and more superior in the sense of computational efficiency.