Artificial Intelligence-Based Approach for Prediction of Power Transformer Health Index

The power transformer is the backbone of the power grid system. A reliable in-service operation of the power transformer is necessary for the efficient availability and distribution of the power. For maintenance of the power transformer, several standard tests are performed periodically to determine the condition (referred to as health index (HI)) of the power transformer. The method of determination of HI requires human involvement, which not only increases the maintenance time but may also cause an error in labeling the HI depending on the experience and understanding of the operator or engineer. In this paper, an artificial intelligence-based approach is employed for the prediction of the power transformer HI using measured data of different transformer tests. The best accuracy of 96.3% was achieved with the decision tree classifier. It is shown that the insulation resistance test has the best correlation with the HI of the transformer.