利用矩阵拉直运算得出了总体最小二乘中随机模型的求解方法,即当某一列向量1等于一变量矩阵与另一列向量2(常数向量)的乘积,且已知变量矩阵的所有元素间的方差及协方差时,确定列向量1的协方差矩阵的方法。推导了倍乘关系、和或差关系、线性、非线性函数关系下以及两个相关变量矩阵之间的误差传播定律,并通过算例说明了该误差传播定律的有效性。
The propagation law of errors in total least squares( TLS) was proposed. The covariance matrix of the first column vector can be determined when the first column vector equals the product of one variable matrix and the second column vector,and the variance and covariance of all the elements in the variable matrix are known. The propagation law of errors between two relevant variable matrixs can be reducted. Some examples indicate that the law is effective.