Two Graphical Shapes Based on DGA for Power Transformer Fault Types Discrimination

Abstract

Different graphical diagnostic methods, which are based on dissolved gas analysis (DGA) are widely used for fault diagnosis and monitoring of dielectric liquid filled transformers. The aforementioned graphical diagnostic methods mainly focus on three, four, five or seven gases. In this work, an alternative approach is proposed by means of a two-shapes graphical method in order to discriminate between different fault types guided from the acetylene percentage derived from the DGA analysis. Low thermal faults are studied through the first shape (square), while electrical discharge faults are depicted in the second shape (pentagon). The accuracy of the proposed method is calculated based on 375 DGA data; the overall obtained accuracy is 78.93%.