Previous work has shown that the normal measurement rate based on measurement uncertainty is a credible evaluation index for state estimation when the true state is unknown.Then a new robust state estimation algorithm based on maximum normal measurement rate is proposed.By approximating the evaluation function，a nonlinearly continuous variable optimal model is developed，and the interior point algorithm is used to solve this model to obtain approximate results.Based on the above study，this paper proposed an improved algorithm by checking the consistency of abnormal measurements to obtain solutions with a higher normal measurement rate，and filtering the noise of normal measurements based on least square estimation with inequality constraints to obtain a more accurate solution.Numerical tests on different systems show the proposed algorithm is efficient for obtaining robust and accurate results，while the increased computing time is acceptable.