In order to more accurately analyze electrical energy loss for the low-voltage distribution network, the feature of “virtual line resistance” is calculated based on the measured data when the topology of the distribution network and the data of a single user’s meter are unknown. Then, combined with the measured power supply and sale, current, and statistical line loss as the features, the isolated forest algorithm is used to detect the data outliers. After removing the abnormal values of the data, the regression analysis method in machine learning is adopted for a large amount of measured data to establish an actual line loss estimation model. The results show that the fitting effect for the actual line loss estimation model based on the features of the measured data is better than based on the non-measured data such as “transformer district information” in previous studies, and provides a reference of the electrical energy loss management for the low-voltage distribution network.
Keywords Low-voltage Distribution Network, Electrical Energy Loss, Measured Data, Actual Line Loss