Mathematical Model of Four Typical Defects for UHF Partial Discharge in GIS

Abstract

Using ultra-high frequency (UHF) method, partial discharge (PD) signals are detected based on four typical physical models of defects in gas insulated substation (GIS). Unified and normalized Lorenz plots are created from three dimensional n-u- t maps of the sampled data. Corresponding mathematical models are established using a five-rank Gaussian functions, and the principles and methods of establishment are summarized. R-square and root mean square error (RMSE) are defined to quantify the fitting error. Furthermore, the normalized correlation coefficient (NCC) and the smooth pseudo Wigner-Ville distribution (SPWVD) analysis demonstrate that there is a strong similarity between the measured UHF PD signals and the computed mathematical representations. Results also show that the UHF PD mathematical model can facilitate theoretical simulation of different insulated defects by using the values of given parameters. In the end, applications of the UHF PD mathematical models and complex wavelet transform for extracting the relevant signals from a white noise background are illustrated; the db3 complex wavelet and six levels of decomposition effectively reveal the relevant signals.