Propagation Characteristics And Energy Distribution Of Partial Discharge Signals In High-Voltage XLPE Cables

110kV and above high-voltage cross-linked polyethylene(XLPE)cables are an important part of power transmission lines.Therefore,the insulation conditions of the high-voltage XLPE cables are of great significance to the safety and stability of the power system.Partial discharge(PD)detection is an effective means for detecting insulation defects in XLPE cables and avoiding further deterioration or even breakdown.Among all the available methods for PD detection,the high-frequency current method has become the most popular one due to its advantages of high sensitivity,strong anti-interference ability and ease of implementation.A high-voltage cable line is usually composed of three single-core cables.In that case,PD signals will propagate in the three-phase cables simultaneously during on-line PD detection and present different three-phase energy distribution characteristics at the detection points.However,existing research on PD detection is mainly concerned with single-phase cables,and there is a lack of systematic analysis and discussion of high-voltage cable lines.For that reason,this thesis delves into the propagation model of PD signals on three-phase cables.Then,the propagation and energy distribution characteristics of the PD signals are fully investigated based on the propagation model.Furthermore,some novel ideas and methods are proposed for PD signal processing and analysis utilizing the energy distribution characteristics.The main results are as follows:The traditional cable distributed parameter model has been improved to overcome its shortcoming in describing the propagation characteristics of the high-frequency components of PD signals.In the improved model,the influence of the semiconductor layer is taken into account,and the equations of the earth parameters with higher available frequencies are provided for directly buried cables and tunnel installed cables,respectively.The contribution of the series impedances and parallel admittances of different XLPE cable layers in the distributed parameter model is studied.Moreover,modal analysis of three-phase cables is carried out by resorting to the modal transformation theory,and the characteristic admittance matrix in the frequency band of the high-frequency current method is derived.Subsequently,the propagation characteristics in different modes are discussed.The results show a large difference between the propagation characteristics of the signals in the tunnel installed cable and the directly buried cable.The earth-return mode and intersheath mode of the former have faster propagation velocities and smaller attenuation coefficients.Finally,the construction of complicated cable networks is realized using the admittance parameters of 2n-port networks,which can be used to obtain the node voltages and node currents together with the node constraints,and thereby provides technical support for the simulation study of high-frequency current detection of PD signals.The energy distribution of the PD signals at the cross-bonding links of three-phase cables is analyzed theoretically.The cross-bonding wires of the cable joints adjacent to the PD source will have PD signals with the same polarity when the wire length is ignored.Moreover,the distribution presents the characteristics of "two large signals and one small signal," and the energy ratio is only determined by the characteristic admittance of the cable line.Then the effect of cable laying parameters on the energy distribution is analyzed using the energy ratio.It is shown that the shorter the cable spacing and the distance from the cable to the tunnel wall are,the greater the differences between the signal energies on the three-phase cross-bonding wires are.In the case where the cross-bonding wires are medium-voltage coaxial cables and the wire length cannot be ignored,different energy distribution characteristics will be found at different detection points of the bonding wires.For the outgoing lines of the cable joint,the signal energy of the defective phase is the largest.For the bonding wires in the cross-bonding box,the signal energies of the three phases are basically the same.Furthermore,simulations and experiments have been conducted and the results are in agreement with the above conclusions.This study is useful for determining the phase of the PD source when performing high-frequency current detection.A three-phase energy ratio method is proposed for separating cable PD pulses and external interference pulses by investigating the energy distribution characteristics of different pulse sources on combined cable-overhead lines.First,the energies of the detected three-phase signals are mapped to a ternary plot to determine the defective phase.For the pulse signals with the same defective phase,the high-frequency and low-frequency energy ratio maps are constructed so that the overlapping cable PD pulses and external pulse interference can be further separated.The feasibility and effectiveness of the three-phase energy ratio method are verified by simulations.A better separation effect is presented in the application on field data in comparison with the equivalent time-frequency method.Compared with traditional waveform-based separation methods,the result of the proposed method is insusceptible to the change of the pulse shapes.Therefore,it can be applied together with the existing separation methods to improve the accuracy of on-line PD detection in cables.An energy estimation method based on noise parameters is proposed to extract the energy of the detected PD pulse signals by reducing the influence of the white noise and narrowband noise in the noisy environment.First,the detected signals are divided into signal frames and noise frames.Then,the 3F-C method is adopted to estimate the noise parameters of the narrowband noise,which can significantly improve the estimation accuracy.By reconstructing and eliminating the narrowband noise,the mixed noise parameter of the white noise and narrowband noise can be obtained.Finally,according to the probability distribution function of the energy spectral coefficients of the field signals,the maximum likelihood estimator of the pulse signal energy is derived,which can be used to calculate the pulse signal energy in the signal frame based on the estimated mixed noise parameters.Applications on both simulated data and measured data show that the proposed method can effectively estimate the energy of PD pulse signals.Compared with the traditional wavelet packet denoising method,the proposed method has a higher accuracy and is less susceptible to the length of the sampling time windows.