Research On Frequency Spectrum Characteristics Of Partial Discharge Audible Sound Signal In Simulating Oil-paper Insulation Defects In Transformer

Oil-immersed power transformer is the main electrical equipment of the power network,and its insulation performance will affect the safe operation of the power network.Due to various uncontrolled factors during production,installation,and operation,the insulation structure may be destroyed,resulting in partial discharge at the position where the insulating medium inside the transformer is damaged.If a fault occurs inside the transformer,it will cause a large-scale blackout in the relevant area,and even cause serious economic losses.By detecting the different spectrum characteristics of the partial discharge audible sound signal of the oil-paper insulation defect inside the transformer in this thesis,the insulation performance and insulation state of the transformer are accurately judged,and the feature extraction method is used to classify the different discharge audible sound signals obtained in the test.Finally,the program of signal aliasing is proposed.The main contents are as follows.(1)According to the common insulation faults of transformer,three kinds of discharge models including pin plate,surface discharge and air gap discharge are designed.By using the existing partial discharge detection methods,an audible signal acquisition system test platform is built to simulate the partial discharge of oil paper insulation defects in transformer.Then,the partial discharge initial voltage and discharge breakdown voltage are obtained through the use of the partial discharge test platform,and the gradually voltage-rising method is adopted to carry out the audible sound signal acquisition test of partial discharge.(2)The test obtains discharge phenomena of three models in the development of partial discharge,and the results show that the discharge phenomena of three discharge models are basically the same.In the early stage of discharge,bubbles,discharge sparks and sound will be produced.During the development of discharge,bubbles gradually increase.Until the late stage of discharge,the weakest part of the insulation paperboard is broken down with obvious black carbonization traces,and bubbles in the insulating oil increase sharply.By collecting voltage,leakage current and audible sound voltage signal simultaneously through three signal channels of oscilloscope,and analyzing the time-domain correlation among them,it is concluded that the time-domain waveform of audible sound voltage signal of partial discharge is bipolar pulse waveform.There is obvious one-to-one correspondence between time-domain waveform of audible sound voltage signal and leakage current signal,and the audible sound voltage signal and leakage current signal all appear near the peak value of the voltage applied to the test.(3)The test gains the audible sound signal of discharge under different discharge types,different discharge gap distances and different voltage levels.The thesis uses the method of Fast Fourier Transform(FFT)and Short-Time Fourier Transform(STFT)to conduct a time-frequency analysis of the audible sound signal of partial discharge in the laboratory.The analysis results present that the main discharge frequency of the audible sound signals of the three types of discharge models are odd-frequency multipliers and even-frequency multipliers of 50 Hz with small vibration amplitude,and the discharge frequency are mainly distributed in low frequency within the range of 0-1000 Hz.But the type of discharge model cannot be distinguished only from the time-frequency spectrogram,therefore,the two methods of voice signal feature extraction,time domain of short-term energy and short-term zero-crossing rate and frequency domain of wavelet packet transform,are used to extract the characteristic parameters concerning the different discharge types of the audible sound signal.Meanwhile,the differences between the characteristic parameters of the audible sound signal of three discharge models are compared and analyzed.As for the results,it can be seen that the short-term energy and the short-term zero-crossing rate of the discharge audible sound signal vary with different discharge models,and the wavelet packet analysis shows that the energy percentage of each frequency band of the discharge audible sound signal is also different.Thus both of these two feature extraction methods can be used to distinguish the discharge types.(4)The test obtains the noise signals of the normal operation of the transformers in Baiyin 330 k V and 750 k V substations,and FFT is used to analyze the time-frequency characteristics.From the analysis of the noise signals,it is known that the frequency of the noise signals near the normal operation transformers is 100 Hz,200Hz and 300 Hz,etc.,and it is mainly distributed in the frequency range of 700 Hz.In order to simulate the discharge signal at the transformer site,the program of signal aliasing is proposed.This program is to mix the audible sound signal of the partial discharge of the simulated oil-paper insulation defect in the laboratory with the on-site noise signal of the transformer,that is,the audible sound signal of the discharge at the frequency of 50 Hz multiplied by the laboratory is mixed with the noise signal of the frequency of 100 Hz in the normal operation of the transformer,and the spectrum characteristics of the aliased signal are collected and analyzed.The aliased signal also has the frequency domain characteristics of the two signals before the aliasing,which verifies the feasibility of signal aliasing scheme.The test verifies that the signal aliasing method can be used to simulate the audible sound signal of discharge at the transformer site,laying the foundation for the subsequent establishment of the audible sound signal database of the transformer discharge.