Research On Coupling Characteristics And Detection Method Of VFTO Signal On Power Transformer Core Earthing Conductor

With the development of power industry,Gas Insulated Switchgear(GIS)has been widely used,and the problems during its application have become increasingly prominent.Especially,the Very Fast Transient Overvoltage(VFTO)caused by switching operating in GIS seriously threatens the safety of the transformer connected to GIS.In the case that the threat can not be completely eliminated,the common method is to timely and reliably detect the VFTO impact suffered on the transformer,effectively record the impact times and strength,which can help to evaluate the insulation and mechanical health status,provide reference for formulating maintenance scheme,and provide data for post-analysis of transformer faults,so as to avoid similar accidents from happening again.During the field practice,the intelligent power station research group found that,VFTO intruding into transformer will couple the corresponding VFTO signal on core earthing conductor,which provides a new way to detect VFTO impulse.Compared with the existing VFTO impulse detection methods,the new way has no influence on the normal operation of the transformer,and has obvious practical significance.Therefore,on the basis of reviewing the relevant research and development of detection technology of VFTO,and summarizing the deficiency of existing Core Earthing Signals(CES)monitoring,combined with the field experience and the actual needs of the engineering,the research of relevant theory and method of VFTO impact detection on the transformer core earthing conductor is carried out.In order to prove the feasibility of this research in theory and lay the foundation for the follow-up method,the whole process of VFTO from generation to coupling to core earthing conductor have been studied.Firstly,extablishing a time-varying parameter model of disconnector,and the generation and characteristics of VFTO are simulated.Then,referring to the existing high-frequency model of winding and combining with its structure,extablishing the winding's "parallel multi-conductor lossless transmission line and lumped parameter hybrid model" to simulate and anlyze the propagation process and characteristics of VFTO on the winding.Finally,according to the core structure and material parameters,extablishing the high-frequency capacitance coupling model of the core,and uniting with the previous model to simulate and analyze the way of VFTO-pulse generated in CES.Then the VFTO-pulse features are summarized and the VFTO-pulse knowledge is established.Research results show that the VFTO invading the winding propagates in two ways of "transmission along the line" and "capacitance coupling".VFTO can couple to core earthing condcutor through the capacitance between the winding and the core,and produce VFTO-pulse in CES.In addition,an actual case is modeled and simulated,the simulation results are basically consistent with the measured results,which proves the correctness of the model and the effectiveness if the simulation.Through this part of the reaearch,the feasibility of this research project is proved,and the knowledge and theoretical basis for the realization of this research goal is also laid.As the basic premise of realizing the goal of this research,the real-time monitoring method of CES is studied.Firstly,according to the time-frequency characteristics of various signals in CES,the appropriate monitoring range and frequency band are selected.On this basis,a front-set unit is designed to carry out “Wide-view” sensing of CES and capture various signals in CES.Then,the post-processing unit is used for “No-blinking” high-speed acquisition and real-time data processing of CES signal,and various signal features in CES are calculated from different angles,so as to extract all the features of CES accuratlly.Finally,the field application cases prove that the proposed CES monitoring method can realize the comprehensive acquisition of signals and the real-time and accurate extraction of features in CES,which provides a necessary method support for the realization of this research goal.On the basis of CES monitoring,the automatic recognition technology of VFTO pulse is studied.Firstly,taking a large number of field pulses as samples set D0,using Relief feature selection method to select 10 features of these pulse,which can best reflect the difference between VFTO pulse and non-VFTO pulse,and constructe the initial sample set D1.Secondly,Principal Component Analysis(PCA)is used to reduce the dimension of D1,a new sample set D2 with only two principal components is constructed,which is used as the training sample for machine learning.Thirdly,for a new sample,the k-Nearest Neighbor(KNN)learning method is used to match its nearest "neighbor" in D2,and the class of the new sample is identified according to its "neighbor".Finally,for the identified VFTO pulses,their trend characteristics are matched with the trend characteristics in the VFTO pulse knowledge base,and their category(closing VFTO pulse and opening VFTO pulse)can be identified.In addition,the test with the field data proves that the proposed VFTO pulse recognition technology can accurately recognize VFTO pulses and identify their categories accurately,which can provide a strong technical support for the realization of this research goal.Based on the research results,the CES online monitoring system is studied based on the Homs collaborative monitoring platform.This system includes four functional layers.The CES data acquisition layer obtains the feature data of CES;the data access layer obtains the power-station-level-data from the HOMS platform and carries out unified management;the data analysis layer carries out multi-source data correlation analysis on the CES feature data and the obtained power-station-level-data,identifies anomalies and auxiliary diagnoses faults;the data service layer provides "prompt and alarm" to users,and provide a variety of interactive interfaces for experts to analyze and diagnose.Meanwhile,considering the complex coupling relationship between the multi-characteristics of transformer and the connected equipment,the multi-source data association analysis technology based on K-L divergence is also studied,and the practical work of transformer multi-state data association analysis is carried out.This system has been successfully applied in Three Gorges,Gezhouba,Xiangjiaba,Geheyan hydropower stations and many substations.Field cases verify the system's ability to detect VFTO impact on transformer,as well as its ability of perception and auxiliary diagnosis for other abnormalities or faults.The cases also show the system's positive role in ensuring the safe and stable operation of transformers.