Statistical approach for interpretation of power transformers frequency response analysis results

Abstract

In spite of many research efforts on fault diagnosis using frequency response analysis (FRA) method, there is still no universally accepted and systematic interpretation technique for these tests and 'expert opinion' is often sought when any damaging trend is observed as a result of fault occurrence. This study deals with statistical criteria for interpretation of the FRA results and presents a number of statistical and mathematical indicators that have not been used so far. A measurement setup accompanied by a transformer as a test object, on which winding faults are imposed, completes the test setup for performing the FRA measurements. The parameters are applied to the experimental FRA measurements and the results are compared by a simple normalising equation. Finally, the indicators which are adding more objectivity and transparency to interpretation of the FRA results are presented. Sweep frequency response analysis (SFRA) method is used in this study because of its efficiency and simplicity. It was found through this research work that F-test, which introduced in this study for the first time, has higher sensitivity, and the differences between the normal and faulty responses of the transformers' windings is more reflected in this parameter compared with the other ones.

Cross Test Comparison in Transformer Windings Frequency Response Analysis

“…They are widely described in the literature [9,[14][15][16][17][18][19][20][21][22], therefore only basic information of their application have been presented in Table 1. Formulas that are designed for assessment in one test setup only, e.g., from Chinese standard DL/T911-2004, which is to be used in end-to-end test setup, have been omitted.…”

Application of DIP to Detect Power Transformers Axial Displacement and Disk Space Variation Using FRA Polar Plot Signature

“…As mentioned in the introduction section, interpretation of FRA magnitude signature based on calculating some statistical parameters such as correlation coefficient (CC) and absolute sum of logarithmic error (ASLE) are not reliable as stated in several publications in the literatures [7][8][9][10][11][12]. To prove this claim, the FRA magnitude signature of the 10kVA transformer, phase-A HV winding is processed to calculate the CC and ASLE based on the following equations:…”

“…While many papers investigating the impact of various winding deformations on the conventional FRA magnitude plot can be found in the literature, no attempt has been made to improve its accuracy to detect incipient and minor deformations [5,7]. A few studies have been conducted in order to ease the interpretation process of the transformer FRA signature through calculating various statistical indicators such as correlation coefficient, standard deviation and the absolute sum of logarithmic error [8,9]. The correlation coefficient however, may lead to a wrong correlation between the two investigated signatures under certain circumstances and is considered as inadequate parameter for FRA interpretation [9].…”

Investigation of the Effect of Winding Clamping Structure on Frequency Response Signature of 11 kV Distribution Transformer

“…Several statistical and artificial intelligence (AI) techniques have been proposed for interpretation of mechanical changes in transformer windings based on FRA. Statistical techniques such as correlation coefficient (CC), absolute sum of logarithmic error (ASLE), minimum-maximum ratio (MM) and absolute average difference (DABS) have been proposed for the interpretation purpose in [3][4][5][6][7][8][9]. The Pearson's correlation coefficient (PCC), Spearman's correlation coefficient (SCC) and Kendall's correlation coefficient (KCC) are compared in [10][11][12] using distributed dataset variables to distinguish the correlation characteristics.…”