An Improved Algorithm of Total Least Squares for Linear Models

Aiming at the condition that among observation equations there are errors in both observation vector and structure matrix,we propose an improved algorithm of total least squares(TLS).Meanwhile,when observation vector and structure matrix are with unequal precision,we deduce the calculation formula using an adjustment model with constraints.Analysis and calculation show that the algorithm that improves approximation accuracy of recursive TLS effectively,can satisfy the conditions needed by fitting equation.