Abstract: To address the under-metering of electrical energy caused by metering faults, this paper proposes a counterfactual electrical energy compensation method based on line loss data and ensemble learning. The method addresses the limitations of traditional electrical energy compensation methods, including low fault localization accuracy, large errors in theoretical line loss prediction, and poor time-of-use disaggregation performance for long-duration faults. It establishes a complete technical framework comprising accurate fault localization, dynamic line loss rate prediction, daily electrical energy compensation calculation, and intelligent time-of-use electrical energy disaggregation. A sliding-window Z-score method is used to automatically detect the start and end times of faults. An XGBoost model is used to predict the dynamic theoretical line loss rate, and SHAP values are introduced to enhance model interpretability. Daily compensated electrical energy is calculated based on counterfactual logic, and an Informer model is introduced to achieve accurate time-of-use electrical energy disaggregation for long-duration faults. Case study results show that the mean absolute percentage error (MAPE) of line loss prediction using the XGBoost model is 4.2%, and the prediction accuracy is 51.7% higher than that of the multiple linear regression model. The time-of-use disaggregation accuracy of the Informer model is 18.3% higher than that of the similar-day method. The proposed method provides technical support for electrical energy compensation under metering faults.