Study On Partial Discharge On-line Monitoring And Location Of XLPE Cable

With the development of electrical industrial, XLPE has been used widely because of its superior electrical and mechanical property. In order to ensure the safety of transmission network, detection of XLPE has been paid great attention, and the on-line monitoring and location of partial discharge occurred in XLPE become one of the issues of general interest.Based on the transmission characteristic of partial discharge in cable, a zCable model considering skin effect is established by using PSCAD/EMTDC. According to simulation results, the transmission and attenuation of partial discharge have been verified which set the foundation for delay-time estimation algorithm for PD sites location.A self-integrating network based broad band current sensor for PD is designed and performance parameter are analyzed. Moreover, the response sensitivity and frequency breadth are studied from integrating resistance, magnetic core material of coil, sectional area of wire and number of windings. Front end circuit is designed and an experimental platform is established.According to specific types of partial discharge of XLPE, 3 typical discharge models are designed. During the experiment, PD signals corresponding to four defects are collected, and noise contained in them are eliminated by wavelet theory. Simulation indicate de-noise has achieve good results.Two time-delay estimation algorithms for XLPE PD sites location, namely, wavelet based general correlation method and twice weighted method, are proposed. Wavelet based general correlation method relax the assumed condition for time-delay estimation of direct correlation method, overcome the inaccuracy of location when PD signal correlate with noise, extend the application of time-delay,and have high degree of accuracy on scale 2~3. Twice-weighted method based on wavelet is not only particularly suit for time-delay estimation under low SNR,but also suit for the PD Location in long cable on scale 2~2 and 2~3 . However, it can not overcome the limitation of non-correlation between signal and noise needed by general correlation method. According to the simulation, both of the two algorithms can suit for PD location of XLPE with high accuracy. In practice, it should use them together to count various noise.