Detection of probability of occurrence, type and severity of faults in transformer using frequency response analysis based numerical indices

Abstract Frequency response analysis (FRA) is mainly used as a tool for detecting mechanical faults (especially displacement and deformation of windings and core) as well as electrical faults (especially short- circuits in windings) of transformers. In the FRA method, the transfer function (TF) of transformer is measured in a wide frequency range and these measurements are compared with the reference TF, to obtain information such as the probability of occurrence, type and severity of the fault; thus, appropriate features must be extracted. In this paper, different transformers are considered, and the transformer TFs are extracted in healthy condition and under different fault conditions such as axial displacement, radial deformation, disc space variation, short-circuits and core deformation. Then, on the basis of statistical and numerical indices, new methods are proposed for fault diagnosis (including detection of occurrence, type and severity of fault). In order to achieve an accurate process and verify the validity of the proposed methods, necessary features are extracted by applying indices to the data obtained from the measurements performed on model transformers. After extracting features, the performance of the proposed methods is evaluated by applying the experimental data obtained from actual transformers. The results show that the proposed methods are more accurate than other well-known methods.