Abstract

Asset Management assists in operating electrical utilities at high performance and low cost. The Power Transformer Health Index (PTHI) is considered a good health condition evaluation and decision-making tool. PTHI is used to prioritize maintenance decisions, drive maintenance strategy, manage failure impact before it occurs, asset lifecycle planning, deferral big capitals, manage spare parts plan, and extend power transformer life. This paper presents the PTHI models’ investigation which was conducted on 4324 transformer records using various Artificial Intelligent Machine Learning (ML) algorithms: Artificial Neural Network (ANN), Support Vector Machine (SVM), Naïve Bayes (NB), Decision Tree (DT), Random Forest (RF) and k-Nearest Neighbours (KNN) in R programming language. Several evaluation metrics present comparable analyses using accuracy, sensitivity, specificity, and F1 score. According to the results, the SVM model was found applicable to local electrical utility transformers' health condition assessment. The paper addressed integrating international best practices and AM into the HI model.