Transformer Fault Identification and Classification of Using Machine Learning

India’s electrical distribution networks have experienced transformer failures, causing significant downtime and power supply interruptions. Prompt repair and monitoring can prevent up to 30% of transformer failures, reducing maintenance costs and downtime. Hence, in recent years, the Indian government is funding to modernize the country’s electrical infrastructure, including smart grid technology, predictive maintenance, and reliability standards. This research study focuses on transformer fault detection and classification by integrating current and voltage sensors. Utilizing machine learning techniques, specifically K-Nearest Neighbor (KNN) and Random Forest (RF) algorithms, the system aims to identify and classify various faults such as short circuit, overvoltage, under voltage, high vibration and as well as normal condition. The sensors provide real-time data, which is used to train the models for accurate fault prediction. KNN leverages proximitybased classification, while RF utilizes an ensemble of decision trees to enhance accuracy. The trained models enable quick and precise identification of transformer faults, contributing to early detection and prevention of potential damage. This integrated approach utilizes Machine Learning (ML) to improve the reliability and efficiency of transformer systems in power distribution networks.