Feature Analysis of Numerical Calculated Data from Sweep Frequency Analysis (SFRA) Traces Using Self Organizing Maps

This paper presents a comprehensive investigation of the self organizing map (som) classification process of good and defective power distribution transformers. three main features were extracted from the numerical calculation method of the sweep frequency response analysis (sfra) signals acquired from the transformers. these features are the input vectors for the som classification. analysis of the results has shown the capability of the features and the som classification method to differentiate between good and defective transformers.