Frequency response analysis in diagnosing transformer winding movements

Diagnosis of transformer winding movements through frequency response analysis, (FRA) relies on correct interpretation of the complex FRA results. Direct and manual comparison between two FRA test results is the current approach used by maintenance engineers and each FRA signature usually contains about 2000 data points. An automatic approach to represent the FRA signature using a transfer function with parameters of the gain, the zeroes and the poles is studied in this paper. The transfer function estimation code is developed in MATLAB and sensitivity study is carried out on the effect of the parameters of the estimated transfer functions.