Analysis On The Electromagnetic Wave Propagation Characteristics Of Partial Discharge And Switchgear Structure Influence On The Time Delay Algorithm

High-voltage switch cabinet is an important electrical equipment of power system,which plays an important role in electricity, such as Closing and breaking power line,line fault protection, monitoring data and so on. The operation reliability directly relatedto the power supply reliability and quality. Partial discharge (PD) as a potential threatdirectly affects the safety of power system. The partial discharge signal containsabundant information of insulation states, the Ultra-High Frequency (UHF) method canpredicate insulation faults and locate by detecting the information, which has importantsignificance to ensure the normal operation of the switchgear, reduce maintenancefrequency, to enhance the operation reliability of the power system. The structure ofswitchgear has reflection, scattering and diffraction effect for electromagnetic wavesthat causes the original signal distortion. The distorted signal can present difficulties inlocation. In order to install the UHF sensor in proper position and accurate location,must have a thorough understanding of propagation characteristics of electromagneticwave of partial discharge and the time delay algorithm influenced by the switchgearstructure. The propagation characteristics of the partial discharge Ultra-High Frequencysignals in switchgear and time delay algorithm as basis of PD location has been studied,which is shown as follows.In order to investigate the propagation characteristics of the partial dischargeUltra-High Frequency (UHF) signals in switchgear and to install the UHF sensor inproper position, this paper presents the finite difference-time domain (FDTD) method tosimulate and investigate the propagation characteristics of PD UHF signals inswitchgear. The simulation results show that the switchgear wall significantlystrengthens the UHF amplitude. The thickness of switchgear wall variability does notinfluence the amplitudes of PD UHF signals. Due to the gap of the switchgearcompartment, the UHF signals can be propagated through the gap. There is a gapcritical dimension to receive PD UHF signals in the switchgear compartment when thePD pulse width is invariant. The insulator supporter has an attenuation effect on theUHF signals. The part of test results has proved the simulation results.Bispectrum time delay estimation is used in switchgear for PD location, howeverwhen the signal to noise ratio(SNR) is low or the peak is submerged in the noise, Timedelay estimation with bispectrum based on wavelet shrinkage is proposed. It extended the assumed condition of bispectrum time delay estimation for signal and noise, andwas capable to estimate time-delay effectively under low signal noise ratio. UsingMATLAB emulate the switchgear partial discharge location, the results show thatbispectrum time delay estimation and time delay estimation with bispectrum based onwavelet shrinkage is both effective for signal to non-noise, however, bispectrum basedon wavelet shrinkage can more accurately calculate the time delay between two partialdischarge signals. The result of the test shows that the time delay calculated bybispectrum estimation improved by the wavelet shrinkage approach is most precise.At last, in order to investigate the time delay estimation method of signal arrival inlocation of partial discharge based on the UHF method, the actual and simplifiedmodels of PD location in the switchgear were simulated by the XFDTD and a series ofthe UHF PD signals were obtained. Six time delay estimation approaches involvinggeneralized correlation method, bispectrum method, etc. were used to calculate the timedelay among the UHF PD signals with and without artificial white noises. The resultsshow that the structure of switchgear and noise will seriously weaken the estimationprecision of the algorithm. Power curve method and cumulative energy method issimple but the error is bigger. The time delay of the UHF PD signals calculated by thegeneral correlation approach is good, but invalid when the signals submerged in thewhite noise. However, the bispectrum estimation approach shows better performancefor the UHF PD signals with and without noises. Moreover, the time delay calculated bybispectrum estimation improved by the wavelet shrinkage approach is the most precise,which is conductive to locate PD source in the switchgear.