Abstract: Traceability method of power outage event based on L-SVM was researched in this paper, to solve the misreporting problem and improve the accuracy of power outage statistics. Power outage events were classified into five types according to their impact, and corresponding vector machines were established. For each vector machine, multiple types of data at multiple measurement points were required to reduce the judgment error caused by the data leakage and misuse. In the case of power-off traceability using vector machine SVM, firstly, the LLE is used to reduce data dimension reduction method, reduce the dimensional disaster of the vector machine and increase the calculation speed, and then establish a combination of five types of vector machines for power outage event traceability, and a full-process analysis method based on L-SVM for power outage tracking is formed. In the end,? Based on the power outage data in the first half of Anhui 110kV Substation, an example is taken to verify the effectiveness of the proposed method in this paper.