In view of the fact that grid topology errors and bad telemetry information affect the safe operation of power grids seriously, this paper proposes a topology identification method based on improved local outlier factor algorithm. The method uses statistical theory to evaluate the state information of breakers and disconnectors and telemetry information of the power network. While taking into account the different influence of telemetry and remote communication on topology error identification, we weight its data by relative entropy. In the process of abnormal topology state detection, non-anomalous objects are shielded to enhance efficiency of the algorithm. Experimental results show that the algorithm can quickly identify the grid topology errors and find bad telemetry information.