Abstract:To address the practical issues of low anomaly detection accuracy and large operation error deviation of low-voltage electric energy meters in operation, combined with the actual situation that low-voltage users only collect daily frozen meter code data,we propose a detection method for online monitoring method of low voltage user electric energy meter operation error based on multi-model fusion.First, we enhance the anomaly detection capability by improving the correlation analysis model with an outlier detection algorithm. Next,we estimate variable losses in the substation using ridge regression to improve the operational error analysis model, overcoming the adverse effects of variable correlation and nonlinear factors on the operational error values of electricity meters. Then, we propose a review model to deeply explore the association between user daily electricity consumption and residential distribution area line losses. Finally, we integrate and comprehensively evaluate the proposed models to achieve precise detection of abnormal low-voltage electricity meters. The case study, based on real data from a province, demonstrates that the method presented herein exhibits high accuracy and superior robustness in anomaly detection and error estimation.