Research On Denoising And Feature Extraction Of Partial Discharge Signal In Power Equipment

Along with our country vigorously construction of UHV ac/dc system,the stability and reliability of the power grid for the construction of the UHV ac/dc system is more and more important.Electrical equipment insulation damage is largely caused by partial discharge,the detection of partial discharge signals of pulse current method and its research on waveform characteristics have great significance to the fault diagnosis of electric equipment.This paper around the partial discharge signals denoising and waveform feature extraction mainly done the following work:Firstly,based on the study of the partial discharge signal denoising,a high voltage filter device with scalable connection partial discharge is designed.The principle of the filter device is briefly described,and briefly introduced the type and structure of the high voltage filter,besides the frequency characteristic is deduced in detail.It is explained that the frequency range of the filter is suitable for the filter section of the partial discharge signal test circuit.The partial discharge signal of the test transformer can be prevented from entering the partial discharge measurement circuit of the tested product and provided technical support for the local test site.Secondly,on the basis of analyzing the insulation structure of the electric equipment and the type of the insulating medium,the four types of typical insulating medium defects are summarized.Four kinds of dielectric partial discharge defect models were fabricated: epoxy dielectric discharge model,air gap discharge model,neon gap discharge model,oil gap discharge model.And the partial discharge test platform was constructed,to carry out the partial discharge test of these four insulating media defects,and the signal was detected by pulsed current method,the partial discharge waveform data was collected by Tektronix DPO 7054 oscilloscope.Thirdly,the waveform feature extraction of four kinds of PD signals is studied,including time-frequency characteristics such as rising edge time,duration,half wave time,frequency center of gravity,frequency range,and so on.For each discharge signal data Time domain characteristics are analyzed statistically,to find out the law and differences;the statistical analysis results as a basis for pattern recognition.Fourthly,on the basis of the theoretical research,involved in the development of a new type of PD,and the time-domain characteristic parameters of the four kinds of partial discharge single-pulse signal waveforms which are analyzed statistically are applied in novel PD,realized the intelligent recognition of the discharge signal and the new waveform map.The accuracy and reliability of the digital PD are verified by the type identification of the PD samples.