부산대학교 도서관

In electrical systems, power transformer failure prediction is a strategic procedure aimed at ensuring the uninterrupted delivery of electricity. In this context, companies of electricity employ predictive maintenance techniques to evaluate the status of this type of equipment. Aiming to contribute by adding monitoring criteria and complementing research in this field, to help maintenance teams project the future status of their power transformers immersed in insulating mineral oil with nominal voltages rated from 34.5 kV to 230 kV and above, with powers from tens to hundreds of MVA, and with and without on-load tap changer, this article presents an innovative method for projection the performance index of these assets on any given date, and to obtain a numerically reliability projection for that date. The originality of this tool stems from its combined use of two easy-to-implement mathematical criteria, considering the historical results of current and previous maintenance dielectric and electrical tests: the Weighted Moving Average and the Consistency of the Historical Neighborhood. Thus, this method is expected to contribute to more accurate to decisions making by maintenance teams.