Alternative statistical techniques for aiding SFRA diagnostics in transformers

In order to aid the interpretative capability of the sweep frequency response analysis (SFRA) technique used for deformation diagnostics of transformers, two new statistical indicators, viz. the t-test and comparative standard deviation (CSD) have been proposed in this study. The motivations behind suggesting the t-test are two-fold, viz. (a) the ambiguity in the results provided by some statistical parameters for borderline cases, and (b) the fact that commonly used statistical parameters require the same sample data size and spot frequencies of the SFRA signatures being compared. The CSD parameter is a logical extension of standard deviation for the comparison of two data sets. The advantages of using the proposed statistical indicators have been demonstrated through six case studies.