Research on Line-transformer-user Topological Anomaly Recognition Model Based on Multi-source Data Mining

The accuracy of the distribution network topology relationship is the basic guarantee for the management of low-voltage and medium-voltage line losses. Due to many historical reasons and economic reforms of the power grid, the line-transformer-user file relationship is abnormal, making on-site investigation difficult. In view of the above problems, combined with the actual scenario of the distribution network, a distribution network topological anomaly recognition model based on multi-source data mining is established. Algorithms such as isolated forest, K-means feature cluster analysis, LSTM model are applied, and multi-dimensional feature fusion is used to identify features of line-transformer-user operation data. The value of intelligent distribution network data is deeply explored. Through the application of examples, the relationship of line-transformer-user in line loss management is realized, effectively improving the efficiency of line loss management.