Research On Partial Discharge Characteristics And Identification Of Typical Defects In Oil-impregnated Paper Capacitance Bushing

Oil-impregnated paper capacitance bushing is widely used in the outgoing line device of power transformer.It connects transformer and external electrical network and plays an insulation and supporting role.It must meet the electrical,thermal,magnetic,mechanical and other requirements during operation.Its insulation state is related to the operation reliability of power grid.This paper obtains the partial discharge(PD)data of typical defects of bushing through experimental research,studies its discharge characteristics,and studies the identification method of PD of typical defects based on multi-dimensional characteristic parameters,so as to improve the efficiency of condition monitoring and operation and maintenance of bushing.In this study,the scaled model of 40.5kV oil-impregnated paper capacitance bushing is taken as the object,and the potential distribution and electric field distribution of the bushing are analyzed to verify the equivalence of scale model by using finite element simulation software.Based on the actual failure reason of the bushing,four typical defects models of the bushing are made,which are the suspended discharge model at the top of the bushing,the suspension discharge model of the pressure sharing ring of the bushing,the surface discharge model of the porcelain sleeve under the bushing and the poor grounding model of the end screen of the bushing.The PD test platform of the bushing is built.The PD signals of each model were collected by pulse current method.The PD signals are analyzed by PRPD spectra and four kinds of two-dimensional spectra,i.e.maximum discharge phase distribution,average discharge phase distribution,discharge repetition rate phase distribution and discharge repetition rate discharge distribution.The PD characteristics of bushing under AC voltage are studied,including the initial PD voltage,phase distribution characteristics of spectrum and discharge distribution characteristics.The results show that the PD of the suspended discharge defect on the top of the bushing occurs in the phase near the peak voltage,the positive half cycle discharge amplitude and discharge quantity are higher than the negative half cycle,the positive and negative half cycle discharge times are close,and the whole spectrum is two segments of suspension;the PD of the suspended discharge defect on the grading ring of the bushing occurs in the phase of the zero crossing point of the AC voltage,and the phase distribution is wider,the distribution of discharge quantity is symmetrical,the number of negative half cycle discharge is higher than that of positive half cycle discharge,and the whole spectrum is three quadrilateral suspension;the PD of surface discharge defect of porcelain sleeve under bushing occurs near the rising edge of voltage,the discharge quantity of negative half cycle discharge is slightly lower than that of positive half cycle discharge,the discharge times are concentrated in positive half cycle,the discharge repetition rate is high,and the whole spectrum is in two triangular shape;the PD of the poor grounding defect at the end of the bushing is the highest,which mainly occurs near the rising edge of the voltage,the positive half cycle discharge is slightly higher than the negative half cycle discharge,the negative half cycle discharge times are higher than the positive half cycle,the discharge repetition rate is high,and the whole spectrum is two-stage suspendedIn order to effectively remove the noise interference in the PD waveform signal and reduce the reduction of the original signal amplitude,an improved wavelet threshold denoising algorithm based on the modulus power threshold function is used to denoising the PD signal.Compared with the other three wavelet threshold denoising algorithms based on the other three threshold functions,the simulation and measured signal denoising results show that this method can suppress the noise well and retain the PD signal characteristics.In order to solve the problems of high dimension and low efficiency of feature extraction in PD pattern recognition,a method of PD feature extraction and recognition for typical bushing defects is proposed,which combines t-SNE and QPSO-SVM algorithm.Firstly,the PD pulse waveform feature,equivalent time-frequency feature,two-dimensional spectrum statistical feature and PRPD image texture feature are extracted,and then the dimension is reduced to three dimension by t-SNE manifold dimension reduction algorithm,and the effect of dimension reduction is visualized.Finally,the typical defects of bushing are identified by QPSO-SVM classification algorithm.Compared with other feature dimension reduction recognition methods,the results show that this method can effectively identify four kinds of bushing defects with high recognition accuracy and short recognition time,which can provide a basis for defect identification of bushing insulation state assessment.