Application of the Fractional Fourier Transform for Detection and Analysis of Partial Discharge Signatures in Distribution Transformers

Several works have been presented about the use of time-frequency representations (TFRs) and signal processing techniques on the study of partial discharges (PD). The fractional Fourier transform (FRFT) is a linear TFR that generalizes of conventional Fourier transform (TF). This paper, presents a study related with the application of the FRFT for the analysis and processing of PD signals generated in distribution transformers. PD signals were acquired using an experimental setup designed and built in the High Voltage Laboratory at the Universidad Distrital Francisco José de Caldas. A total of 26 PD signatures were analyzed in the experiment. Simulation results show that spectrum of PD signatures, computed with FRFT, have a main peak, one or two internal peaks and a set of secondary oscillations (side lobes) associated with high frequency components, which are distinctive of this type of electromagnetic transients. In addition, a statistical analysis of several parameters of the PD signal transform in the FRFT domain is presented.

Several works have been presented about the use of time-frequency representations (TFRs) and signal processing techniques on the study of partial discharges (PD). The fractional Fourier transform (FRFT) is a linear TFR that generalizes of conventional Fourier transform (TF). This paper, presents a study related with the application of the FRFT for the analysis and processing of PD signals generated in distribution transformers. PD signals were acquired using an experimental setup designed and built in the High Voltage Laboratory at the Universidad Distrital Francisco José de Caldas. A total of 26 PD signatures were analyzed in the experiment. Simulation results show that spectrum of PD signatures, computed with FRFT, have a main peak, one or two internal peaks and a set of secondary oscillations (side lobes) associated with high frequency components, which are distinctive of this type of electromagnetic transients. In addition, a statistical analysis of several parameters of the PD signal transform in the FRFT domain is presented.