Feature Selection for Effective Health Index Diagnoses of Power Transformers

Abstract

This letter investigates an approach based on feature selection and classification techniques to reduce assessment complexities of power transformers. This approach decreases the number of features by extracting the most influential ones when determining the transformers health index (HI). Several filters and wrapper-based feature selection methods are investigated. The effectiveness of the selected features is validated through performance evaluations of various classification models. The experimental results demonstrate that water content, acidity, breakdown voltage, and FFA (Furan), are the most influential testing parameters in determining the transformer HI.