Research On State Perception And Fault Diagnosis Method Of Transformer On-load Tap-changer

On-load tap-changer changes the turn ratio of transformer winding without interrupting power supply,and undertakes the voltage regulation task of high current load in power grid.Its reliability is directly related to the safe and stable operation of power grid.The on load tap changer control circuit,energy storage mechanism and change-over switch cooperate closely to perform energy storage,contact opening and closing actions,study the variation characteristics of electric vibration signal in the process of energy transmission,timely find the operation defects of OLTC equipment and quickly locate mechanical faults,which is of great significance to the operation and maintenance of on-site transformer.Based on the analysis of the working principle and common faults of on load tap changer,this paper makes an in-depth study on OLTC state analysis and fault diagnosis by using non-invasive vibration signal processing technology.Firstly,the selection of on load tap changer stat e sensing method is analyzed from the electrical and mechanical characteristics.The electromagnetic transient simulation model is used to analyze the changes of breaking current and arc characteristics under the abnormal state of OLTC contact resistance i ncrease and timing lag.A fault simulation test platform is built,and the relationship between mechanical characteristics and OLTC operation state in the switching process is described.Through comprehensive consideration of the comprehensiveness of describing the state information and the complexity of obtaining the signal,in order to select the information that meets the state perception requirements and characterizes the operation characteristics of the tap changer from the signals with complex component sources.Secondly,taking OLTC body vibration and motor current signal as the research object,this paper focuses on the mechanical state identification algorithm based on the characteristic analysis of electric vibration signal.The envelope contour characteristics of current signal and the texture characteristics of vibration signal are studied.The Hilbert envelope purified by adaptive white noise empirical mode decomposition is segmented,and the key coordinate points of local maximum in each stage are taken as the characteristics;Based on recursive quantitative analysis,the behavior law in the high-dimensional phase point trajectory graph is displayed in the two-dimensional recursive graph.Then,taking image processing as the starting point,the 8-dimensional texture features describing the relationship between pixel regions are extracted by gray level co-occurrence matrix.Finally,the OLTC fault diagnosis method based on random forest algorithm is studied,and an improved algorithm of weighted in tegrated voting with AUC value as index is proposed.The feasibility of the OLTC state perception and diagnosis algorithm proposed in this paper is verified by simulation experiments.