The analysis of abnormal line loss is of great significance in the development of a low-voltage distribution network. At present, the identification of abnormal line loss mostly adopts the threshold analysis method, which has great limitations in timeliness and accuracy. Given the increasing popularity of the smart grid, an abnormal line loss identification scheme based on isolation forest outlier detection algorithm is proposed. First, the k-means algorithm is adopted to cluster the low-voltage station areas according to different load conditions. Then the isolation forest algorithm is employed to calculate the anomaly scores of the station area data. Finally, threshold analysis is performed on the anomaly scores to obtain the abnormal line loss data. A simulation analysis on the IEEE standard distribution network and verification on a real operating grid show that the abnormality identification algorithm proposed in this paper has high accuracy. This abnormal analysis method based on data mining technology will play an increasingly important role in the management of line loss. This work is supported by the National Natural Science Foundation of China (No. 61873194).