Data Analysis Methodology Utilizing the Statistical Metrics Weight of Evidence (WoE) and Information Value (IV) to Assist in Asset Management of Power Transformers

This article proposes an original contribution to assist in evaluating the condition of power transformers, the main asset of the electrical energy transmission sector. Due to constructive similarity, reactors are also evaluated. Data from the insulating oil moisture of the equipment and the following categorical variables were used: voltage class, installation region (regional), criticality, type, and age of the equipments. The feasibility of applying the methodology stands out, since these categories are technical registration data of transformers and reactors and the water content is an essential characteristic for determining the operational condition of the insulating oil, being one of the measured properties in the physical-chemical tests carried out on the oil. The main contribution presented is the selection of categories with higher weight and categorical variables with higher predictive power to direct maintenance actions. This goal was achieved using the statistical metrics Weight of Evidence (WoE) and Information Value (IV). The methodology was applied to a dataset of nearly 10 thousand samples of oil from 795 power transformers and reactors from the ISA CTEEP park, an electric power utility company in Brazil.