Fault Prediction and Classification in Power Distribution Transformer Using Machine Learning

This article used Random forest (RF) classifier, decision tree (DT) classifier, K-nearest neighbors (KNN) classifiers, extra tree (ET) classifier and extreme gradient boosting (XGB) classifier for fault prediction and classification in power distribution transformer to predict the phase line voltages, line currents, line-to-line voltages, neutral current, ambient temperature indicator, oil temperature indicator, oil level indicator, and various other alarm indicators like oil temperature indicator trip oil temperature indicator alarm and magnetic oil guage alarm. The findings demonstrate that XGB and ET classifier, provide superior predictive capabilities compared to other methods.