Enhancing Power Transformer Reliability through Machine Learning-Based Fault Prediction Using Dissolved Gas Analysis

The global energy sector operates within a highly competitive market, necessitating uninterrupted power supply to industrial, commercial, and domestic sectors. Transformers serve a critical role in electricity transmission, emphasizing the importance of maintaining their performance to minimize losses. Dissolved Gas Analysis (DGA) emerges as a pivotal tool for monitoring transformer performance and identifying fault types in oil-immersed transformers. Despite the existence of several conventional DGA interpretation techniques, their accuracy has been subpar. This research paper presents a novel machine learning(ML) approach for predicting power transformer faults using DGA data. The proposed method leverages DGA samples sourced from the IEEE data port to train and test various machine learning models within the WEKA platform.