Study On Interferences Elimination And Fault Diagnostic Techniques For UHF Partial Discharge Detection

In order to make an effective application of UHF PD measurement,several relative research works have been done in this thesis,which includes the interference elimination,PD pattern recognitions and discharge stages evaluation.The main achievements are as follows:The characteristics of various interference signals in UHF partial discharge detection are deeply analyzed.Inside-outside signal antithesis method,inter-phase antithesis method, signal multi-cycle analysis method and clustering analysis method based on grid and density clustering method and fuzzy theory are proposed to exclude external and random pulse type interferences.The effectiveness of these methods is verified by simulation and analysis treatment of large amounts on-site data.Waveform characteristics of four typical UHF interference signals are summarized to provide the basis for interference exclusion.For single ultra-high-frequency partial discharge signals,a signal separation method is proposed in this paper based on time-frequency analysis and cluster analysis method to solve the problem of separation of different types of partial discharge UHF signal,laid the foundation for research of PD pattern recognition.The characteristics and features of the discharge fingerprint of five typical partial discharge fault defects are deeply studied.A method of discharge pattern features optimal selection based on the difference degree calculating is proposed and a variety of discharge types of discharge patterns characteristic parameters are analyzed.Compared with other characteristic parameters,it is shown that characteristic parameters of partial discharge gray image is more conducive to the pattern recognition and discharge stage evaluation through the difference degree analysis of statistical characteristic parameters of discharge patterns spectrum,amplitude characteristic parameters,as well as gray image characteristic parameters.The similarity calculation and gray assessment are used to identify the discharge type and discharge stage.The results show that similarity calculation on partial discharge gray image is more simple and effective for partial discharge pattern recognition and discharge stage evaluation.A set of PD diagnostic software is developed based on our results.It is shown that the software is stable and easy to use through laboratory and field applications.