关于总体最小二乘方法适应性实验研究

摘要

The total least-square has been a more rigorous adjustment method than least-square in recent years,which can take into account and correct the accidental errors both in coefficient matrix and observed values matrix. However,the applicability of the total least-square and its ability of correcting the errors in coefficient matrix and observed values matrix in the model based on actual data,remain not in-depth study. It presents the sensitivity of the total least-square method in one-dimensional liner regression model, and verifies the correction capabilities and advantages in the experiment with simulation data.%总体最小二乘是近年来发展起来的较最小二乘方法更为严密的平差方法,总体最小二乘能够顾及系数矩阵和观测值矩阵同时存在偶然误差并加以改正.然而对于总体最小二乘方法的适用性以及在根据实际数据建立模型时总体最小二乘方法改正系数矩阵和观测值矩阵误差的能力问题还没有深入研究,针对一元线性回归模型,讨论总体最小二乘方法的灵敏性,利用仿真实验数据验证总体最小二乘方法在线性回归模型中的改正能力和优越性.