The weighted least squares（WLS） estimator is the currently most widely used method for power system state estimation. However， the estimation results of WLS could be far from the true values of state variables when bad data are present. To solve the problem， a non-quadratic objective estimator， the weighted least absolute value（WLAV） estimator， and least mean squares（LMS） estimator are proposed. Numerical simulations， including residual pollution examples， are carried out to compare existing estimators and the maximum normal rate estimator（MNR） described in the second paper of this series. Test results show that the proposed method is as good as WLS and better than other methods when no bad data is contained in the test system， and it is far superior to all other methods when test points contain bad data， showing that the proposed method can deal with the residual pollution scenario fairly well.