With the integration of distributed energy resources, the load characteristics of power grids have changed. In particular, under conditions where renewable energy sources such as photovoltaic and wind power exhibit large fluctuations, the grid may experience light-load conditions. To address the misjudgment of wiring errors in electric energy meters under scenarios of light load and reactive power over-compensation caused by a high penetration of distributed energy resources, an intelligent identification method based on XGBoost is proposed. By analyzing features such as current and power factor in light-load and reactive power overcompensation scenarios, and by combining actual collected data with generated data to build a classification model, the method effectively distinguishes normal operating states from wiring errors. In validation tests, the model achieved an accuracy of over 98% in detecting wiring errors, as demonstrated by multiple evaluation metrics, including accuracy, precision, recall, and F1-score, significantly reducing the false judgment rate. The results show that this method can effectively handle complex power load scenarios caused by the high penetration of distributed energy resources, providing an efficient and reliable intelligent solution for identifying wiring errors in power systems.