Research On Feature Extraction Of Partial Discharge Signal Based On Envelope Detection

The phenomenon of partial discharge(PD)produced by high voltage electrical equipment is not only the symptom and manifestation of insulation deterioration,but also the cause of further deterioration of insulation.In order to ensure the safe and reliable operation of high-voltage electrical equipment,it is necessary to conduct real-time monitoring on the degradation degree of insulation defects of high-voltage electrical equipment on site,so as to avoid serious accidents caused by insulation breakdown.Among many PD detection methods,ultra high frequency(UHF)detection is one of the most widely used and most mature methods.However,from the practical engineering application of UHF detection method,it can be found that there are a large number of electromagnetic interference signals in UHF frequency band in the field environment,which will seriously affect the accuracy and effectiveness of UHF detection method for PD signal detection.There is a large number of electromagnetic interference signals in UHF detection frequency band in PD detection field,which will reduce the accuracy of PD signal detection.In order to solve this problem,this thesis uses pulse waveform characteristics phase sequence detection method to replace the traditional pulse peak characteristics phase sequence detection method.In this method,the interference signals and the signals of different partial discharge sources are separated by extracting the effective envelope waveform features,which lays a foundation for improving the accuracy of pattern recognition of phase resolved partial discharge(PRPD)spectrum.Therefore,this thesis designs and implements a UHF envelope detection hardware circuit for UHF PD signal.The UHF envelope detection circuit is used to envelope the UHF PD signal.Based on this,the influence of different propagation paths on the envelope waveform characteristics is studied.By simultaneous interpreting the envelope waveform characteristics,we first select and propose six envelope waveform feature quantities.In order to intuitively compare the effect of these features on the clustering and separation of PD signals,a two-dimensional mapping scatter diagram of envelope waveform features is drawn through software simulation.The simulation results show that the clustering and separation effect of UHF PD signal envelope waveform is better by using two features,amplitude variance and integral area / half wave pulse width.In order to further test the clustering and separation effect of the selected amplitude variance and the integral area / half wave pulse width on the partial discharge signal on the partial discharge detection system,two partial discharge power sources are placed in the laboratory to discharge at the same time.Firstly,using the classical method which is widely used in the field,the PRPD spectrum is drawn directly according to the peak characteristics of the pulse;secondly,the PRPD spectrum of each subcategory is drawn after the clustering and separation of the envelope waveform characteristics.Finally,through the comparative analysis of the two PRPD spectrums before and after,after the time-domain waveform feature clustering separation,the two PD signals which were originally mixed into a group on the PRPD spectrums is successfully separated,and they were clustered on the respective PRPD spectrums,which further verified the effectiveness of the selected envelope waveform feature on the PD signal clustering separation.