A new technique for a better sweep frequency response analysis interpretation

This paper presents a new technique for a better and faster sweep frequency response analysis (SFRA) results´ interpretation. The proposed technique is based on the time-frequency analysis. Unlike the Fourier transform (FT) which is a 1-dimensional representation, a time-frequency distribution (TFD) is a 2-dimensional representation where both time and frequency are present in one plot. Moreover, noise can affect the SFRA signal and affect the interpretation of the results which may lead to misinterpretation of the results. However, noise will have less effect on the TFD signal. By just looking at the TFD of the SFRA signal, we can conclude whether the signal comes from a healthy or unhealthy transformer. To quantitatively assess the strength of the proposed technique, we evaluate the cross-correlation coefficient of the computed TFDs. Different case studies are presented to show the superiority of the proposed technique compared to the existing FT based SFRA.