Transformer health index monitoring using supervised prediction model

Dissolve gas analysis (DGA) is a method used to distinguish between transformers that are in optimal condition and those that require scheduled maintenance. The primary objective of DGA is to accurately identify issues caused by different gas forms in the transformer. The key gas method (KGM) analysis is a frequently employed approach in DGA. KGM is utilised to classify the health index of the transformer based on the development of gases within the transformer. Additionally, the prediction models used for health index classification include K-Nearest Neighbours (KNN), Discriminant Analysis, Principal Component Analysis (PCA), and Decision Tree as decision-making tools. The resulting outcome is subsequently compared to alternative prediction models to determine the ideal performance based on accuracy, precision and recall prediction. The results demonstrate that the KNN prediction model surpasses other models with an accuracy of 94.27%, precision of 94.12% and recall 92.44%.