The concepts of inconsistent outliers and consistent outliers based on the characterization analysis are proposed from the perspective of consistency. Besides

it is pointed out that the possible outlier sets should contain inconsistent outliers and consistent outliers. In addition

a two-step method for state identification is put forward. The method firstly identifies all possible inconsistent outliers and then classifies the possible consistent outliers. For inconsistent outliers

a hierarchical search algorithm is proposed according to the multi-level structure of power system. By localizing the search procedure

the search space is reduced and the efficiency of the search algorithm is improved significantly. For consistent outliers

a sensitivity analysis method is proposed to find the outliers which have the most impact on the result sets. By sensitivity analysis

the combination explosion problem of searching all the active measurements is avoided.This work is supported by National Science Foundation of China(No. 51207136).