Transformer Winding Deformation Diagnosis Algorithm Based On Fusion Of Multiple Indexes

With the rapid development of power grid,the requirements for safe and stable operation are increasing.As a key part of the power system network,the monitoring and diagnosing of the power transformer are critical to reducing maintenance and improving the equipment reliability.Statistics showed that the winding deformation is the most common form of transformer fault.Due to the concealment and cumulative effects of winding deformation,the failure initially had no obvious effect on the normal operation of the equipment.However,when the degree of deformation is deepened,it is very likely to cause irreparable damage,which will lead to a serious power accident.Therefore,the diagnosis of winding deformation is always the core research content in the transformer fault diagnosis.Firstly,in this paper,the relative change values of the windings’ internal inductance and capacitance of the normal displacement and axial,radial and mixing deformation of the transformer’s winding were obtained by finite element simulation.The change law of equivalent circuit parameters caused by the different types and degrees deformation of the transformer windings was summarized.Based on these parameters,equivalent circuits were established and frequency response curves were obtained according to the changed circuit parameters.The variation rules of the frequency response curves’ resonance frequency and amplitude before and after the winding deformation were summarized.Secondly,some numerical indexes to evaluate the frequency response curve of the transformer winding deformation were studied in this paper,the results show that Euclidean Distance is more linear than other numerical indexes under winding axial displacement,which could be used to effectively determine the degree of axial displacement.For distinguishing the type of winding deformation fault,the ED of the different frequency bands in the frequency response curve could effectively show the difference between axial displacement and radial deformation.So the ratio of regional Euclidean Distance could be used as a feature to determine the type of deformation.For the mixed deformation failure of the winding,according to the waveform characteristics of the frequency response curve,the offset area is used to calculate the offset degree of the frequency response curve.The offset area ratio after splitting frequency band could distinguish a single radial deformation and mixed failure.In the end,this paper proposed a transformation deformation diagnosis method that combined the ratio of RED and the ratio of the offset area.These are two layers of this method.In the first layer,the ratio of RED is used to judging axis or radial deformation.In the second layer,the ratio of the offset area was used to distinguishing single or mixed faults in the winding.The fusion index method proposed in this paper could effectively overcome the insufficient ability of the single index when the mixed deformation happened in the winding.