Detection of Abnormal Line Loss Rate in Low-voltage Transformer District Based on VAE

Line loss rate is an important indicator of the safe, stable, and efficient operation of the power system. As the number of low-voltage transformer district at the end of the power supply is huge and the lines are complex, the power loss generated should be paid more attention to. Starting from the low-voltage transformer district, this paper proposes a line loss rate anomaly detection model based on a variational autoencoder. Use the random matrix theory to analyze the correlation of the line loss data, filter out the line loss rate influencing factors, and construct the low-voltage transformer district line loss rate influencing factor index system. On this basis, a line loss rate anomaly detection model based on a variational autoencoder is established, the input features are modeled in the hidden space, and anomalous features are sampled from them. The reconstruction probability of the reconstructed data will be the same as the threshold. It compares the identification of outliers, and proves the superiority of the variational autoencoder in the identification of abnormal data of online damage through the test and analysis of the sample transformer district.