用实验方法确定经验公式时，回归方程中用最小二乘方法求得的待估参数的估计值往往是一个很长的数。
When determining empirical formula with test method, estimation value of estimated parameter in regression equation which is obtained with least square method always is a long number.
总体最小二乘方法(TLS)是近年来出现的一种较新的平差方法，与传统最小二乘方法(LS)不同的是，TLS可以同时顾及系数矩阵及观测值的误差。
The total least squares (TLS) are a new method in adjustment in recent years. Different from the least squares (LS), the errors in coefficient matrix and observations are considered simultaneously.
应用推荐