AI-Based Power Transformer Failure Prediction Using Iot and Machine Learning: A Review

Power transformers are essential parts of electrical power systems. Any failure in these transformers can cause power interruptions, equipment damage, and increased maintenance expenses. This paper reviews selected research on trans-former monitoring, fault detection, machine learning, explainable AI, and remaining useful life estimation. The re-viewed studies use different techniques, including IoT sensors, dissolved gas analysis (DGA), LightGBM, SHAP, LIME, and thermal modelling. These methods help monitor transformer conditions, identify faults, estimate their severity, and study equipment ageing. Although the reported results are encouraging, a direct comparison is difficult because the studies use different datasets and evaluation methods. The review also highlights issues related to data quality, model testing, fault identification, and practical use. Finally, it suggests possible improvements for future research.