The least deviation estimation method of maximum normal rate based on the minimum deviation criterion considering the normal rate of measuring point is proposed. A large number of examples are tested to make a comparison among the method proposed above and the existing methods of weighted least squares

weighted minimum absolute value

non-quadratic criterion and estimation method based on measurement uncertainty. The results show that the method proposed has a strong ability to suppress the bad data with strong correlation and bad data of leverage. The computational efficiency can also meet the requirements of large-scale systems.