Research on Fault Prediction Model of Gateway Metering Transformer Based on Data Mining

Gate measuring transformers, including voltage and current transformers, are critical for accurate electric energy measurement and maintaining grid stability. However, complex operating environments and aging equipment often lead to failures, causing metering errors and potential power accidents. Traditional fault detection methods are inefficient, reactive, and underutilize available data. This paper presents a fault prediction model for gateway metering transformers based on data mining, utilizing random forest algorithms to analyze operational data. Preprocessing techniques such as missing value handling and feature selection with XGBoost were applied. Experimental results show a 97.5% prediction accuracy, significantly enhancing fault prevention and grid reliability.