LAUSR

Partial discharges (PD) are one of the most important factors that affect the insulation of power and distribution transformers. Because their repeated occurrence can produce isolation failures, the identification and analysis of PD pulses are important to guarantee the correct operation of equipment. In the last two decades, some processing tools based on time domain, frequency domain or hybrid domains have been developed to detect and analyze PD signals. Fourier transform, short-time Fourier transform, wavelet transform and Gabor transform are some examples of techniques used to provide information about the amplitude levels in time-frequency bands and identify the power spectrum of off-line/on-site PD waveforms. However, there are currently no proposed techniques to evaluate the performance of the instantaneous frequency (IF) of PD signals and its variation with respect to time. This paper presents the application of the local polynomial Fourier transform (LPFT) and the computation of local polynomial periodogram (LPP) for time-frequency analysis of PD pulses generated in distribution transformers. PD signals were acquired from an experimental setup specially designed and built for the research purposes. To acquire electrical signals, a fast current transformer is used to measure directly the transient current pulses generated by PD. Simulation results show that power spectrum of practical PD signals, computed with LPFT, reveals a low and medium frequency content related to secondary peaks and other high frequency components produced by noise and interferences, which are not detected when the STFT is used.