Research On Frequency Response Modeling And Fault Diagnosis For Large Transformer Winding In High Speed Railway

In recent years,the high-speed railway has become one of the most important means of transportation in China by the advantages of high speed,comfort and strong transport capacity.The traction power supply system provide power and ensure the fast and stable operation of high-speed train.Autotransformer(AT)power supply system is the major power supply mode of high-speed railway with the advantages of small traction network impedance and long power supply distance.And the traction transformer and autotransformer are one of the most expensive and largest quantity equipments in AT power supply system.Traction power supply system are usually threatened by short circuit fault,direct lightning overvoltage and resonant overvoltage,which will cause tremendous forces in transformer windings,even lead to winding deformation or damage.Frequency response analysis(FRA)method is the most widely used method for power transformer winding deformation diagnosis.Affected by the difference of transformer type and winding structure,the frequency response characteristics of traction transformer,autotransformer and power transformer are different significantly.And the transformers in high-speed railway require higher accuracy for winding deformation diagnosis.As a result,Applying power transformer standard for traction transformers and autotransformers directly lack of pertinence and effectiveness.Therefore,it is of great significance for ensuring the safe and reliable operation of high-speed railway by proposeing a pertinent and reliable diagnosis standard for traction transformers and autotransformers.Therefore,this thesis foused on the traction transformer and autotransformer in high-speed railway.According to the research route of “fault feature analysis—test platform construction—frequency response modeling—feature extraction—fault diagnosis method”,the winding deformation fault diagnosis of large transformer in high-speed railway is carried out.The thesis consists of the following contents.1)According to FEA and field test,the short-circuit stress and failure modes of the radial split windings of V/X connection traction transformer and autotransformer are calculated.Then exploring the deviation characterist of major parameters including distributed inductance and capacitance.At the same time,a transformer test platform is designed for simulating various winding faults.The FRA tests are conducted under different winding fault modes and the results are analyzed.It is concluded that the winding connection method of transformer is one of the major factors affecting the frequency response curve,and the sensitivity of different windings to faults is quite different.2)Based on FEA and state space model,an improved lumped parameter model considering the distributed capacitances between all disks for radial split windings is established,which improves th simulation efficiency compared with circuit simulation software.Then a method for calculating frequency dependent inductance considering measured curves is proposed,which improves the accuracy of critical resonant frequency calculation.3)According to the established frequency response model,the frequency response characteristics of traction transformer and autotransformer windings under axial displacement and radial deformation are analyzed.The results show that the frequency response curve of three-winding transformer is quite different from that of double-winding transformer.And critical resonances are proposed to diagnose the faults of V/X connected windings and radial splitting windings.4)Combining the phase-frequency curve and image processing method,the frequency band division method is first proposed.Then two new indices named area ratio(AR)and centroid deviation(CD)are proposed to quantify the deviation of frequency response curves.The AR difference between different frequency bands and whole frequency band can determine the major fault frequency band.And the CD is adopted to quantify the deviation of frequency response curves.Finally,the proposed indices are used to distinguish faulty winding and fault type in autotransformer effectively.5)Based on image features and SVM,the identification of winding fault types and fault locations for the winding connections of traction and autotransformer is studied.The axial displacement,short-circuit and deviation of series capacitance fault of windings are simulated in test platform.The AR and CD are extracted based on the proposed frequency band division method.The image features and the validity of the proposed features in identifying winding fault types and fault locations are verified.The validity of the proposed features for identifying winding fault types and fault locations is verified.