Power Transformer Diagnosis Using FRA and Fuzzy Sets

Abstract

This paper presents a novel method for failure detection and interpretation in power transformers using Frequency Response Analysis (FRA). Mathematical indexes are used for failure detection. The performance of the most used mathematical indexes is evaluated and the most sensitive for frequency response analysis is selected. An approach based on a suitable codification of response frequency bands is proposed for failure classification. A novel characteristic of this work is the application of Fuzzy Sets to represent the mathematical index limits for failure detection. The proposed method has been validated using two different case studies, and the diagnosis results obtained for them show a good agreement with the real failures.