Study of Partial Discharge Based on Time-Frequency Analysis Using Local Polynomial Fourier Transform

In recent years, for partial discharge (PD) analysis signal processing techniques have been applied in the time-frequency domain as Short Time Fourier Transform (STFT), Wavelet Transform (WT) and Wigner Distribution (WD), among others. Local Polynomial Fourier Transform (LPFT) is a linear time-frequency representation which is a generalization of STFT when m = 1. The LPFT has been used in applications, such as: analysis of time-varying signal, radar, sonar and communications. This paper present a study of partial discharge based on time-frequency analysis using Local Polynomial Fourier Transform. The analyzed PD signals were simulated using two mathematical models: PD single-pulse model and PD pulse sequence model. As part of the study, the waveform of PD signals selected were characterize in the time domain and the Fast Fourier Transform (FFT) was used in the analysis at frequency domain. Finally, to demonstrate the versatility of the proposed signal processing technique (based on LPFT) and its application on PD signals, the results obtained with the LPFT of polinomial order m=2  were compared with those obtained applying STFT. The results show that the LPFT is able to reveal low and medium frequency components which are due to the secondary peaks and oscillations of the PD pulses, which are not detected when the STFT is used.