二元线性回归的不确定度评定方法

Method for estimating uncertainties about two variable linear regression

讨论了根据最小二乘法原理拟合二元线性回归方程 y=a0 +a1x1+a2 x2 时 ,系数 a0 ,a1,a2 的估算值及其不确定度评价 ,解决了目前在进行 a0 ,a1,a2 的误差估算时不考虑自变量的误差对结果的影响及不考虑因变量 y的 B分量误差的影响这两大问题 ,提出了更科学更合理的新方法 。

In this article, the estimating of the parameters a 0,a 1,a 2 of the two variable linear regression equation y=a 0+a 1x 1+a 2x 2 according to the principle of the least square fit for their uncertainties is discussed. we try to solve the two big problems presents in the prevail methods used to estimate the error of those parameters:without considering the effect of the error of the independent variables or the B component of the error of the variable y. Finally a new method more scientific and more reasonable is generalized.