Transformer Iron-core State Detection And Diagnosis Method Based On Vibration Signal

Power transformers are primary equipments in the power system,and its safety and reliability are crucial to the reliable operation of the power grid.The vibration of the transformer is closely related to the state of the iron core and windings.In recent years,vibration detection has become one of the main means of comprehensive monitoring of large power transformers,especially ultra-high voltage transformers.However,with the continuous generation of a large amount of vibration monitoring data,how to accurately diagnose transformer faults through vibration monitoring has become an urgent problem to be solved.There have been many researches on detecting transformer core faults using vibration signals,but the current research on core fault vibration is mostly limited to a single fault type,and there is a lack of comparative research on different types of faults.At the same time,for a long time,on-site transformer vibration data is mostly obtained through short-term detection,lacking long-term data characteristics and laws under different operating conditions,and it is difficult to provide support for transformer online monitoring and status assessment.Therefore,based on the investigation and theoretical analysis of transformer core vibration literature,this paper builds models of various typical defects such as loose core clips,multi-point grounding,overexcitation,DC bias,etc.,and experimentally studies the vibration signals of various defects.It lays a test foundation for the analysis and diagnosis of on-site vibration monitoring data;through the analysis of long-term vibration data of on-site transformers,the statistical characteristics of on-site transformer vibration data are obtained.First of all,this paper investigates and analyzes the vibration mechanism of the transformer,and studies the influence mechanism of harmonics,iron core clamping degree and temperature rise and other factors on the iron core vibration,and then proposes the effects of overexcitation,DC bias,loose iron core clamping and excessive iron core.Defect vibration signals such as point grounding are studied.Secondly,using a 10kV distribution transformer as a test product,the above four kinds of defects are experimentally simulated,and the characteristics of the amplitude,odd-even harmonic ratio,high-frequency energy ratio and spectral complexity of the vibration signal under different defects are studied experimentally..The experiment found that,except for the multi-point grounding of the iron core,the characteristic quantity of each defect will increase compared with the normal time,and each has its own characteristics.When overexcited,the amplitude,high-frequency energy ratio and spectral complexity increase most significantly;when DC bias is applied,the odd-even harmonic ratio increases most significantly;when the core clamp is loose,the high-frequency energy ratio and spectral complexity increase most significantly.The test support is established for the state detection and diagnosis of the transformer core.Third,for the vibration monitoring equipment installed on 8 500kV AC main transformers by the research group,the vibration data with a time span of 6 months was collected,and the statistical characteristics of the aforementioned four characteristic quantities were analyzed.The relationship between the remaining feature quantities forms a two-dimensional distribution and uses this distribution to analyze the vibration data.Through the horizontal and vertical comparison,it is found that the distribution of characteristic quantities is obviously outlier compared to the distribution range of most of the other transformers,and it is suspected that there is abnormal transformer vibration data.foundation.Finally,in the comparison of vibration data of longitudinal transformers in different time periods,the regression model is used to fit the two-dimensional distribution of transformers in different time periods into a curve and the average deviation is used to quantify the difference between the curves of different time periods;in the comparison of vibration data of multiple transformers in the horizontal direction On the one hand,the data suspected to be abnormal is eliminated,and the remaining data is statistically analyzed.Combined with the field data analysis results,the horizontal and vertical normal thresholds of the vibration characteristics of the two channels are divided respectively.Finally,a transformer abnormal state diagnosis method is formed,which judges whether the vibration signal is abnormal or not through the horizontal and vertical thresholds,and judges the abnormal type by combining the characteristics of each defect vibration signal obtained by the experiment.Using this method,it is found that the two-channel data of No.6 transformer is over-limited in both vertical and horizontal directions,and the change characteristics of characteristic quantities are similar to DC bias,so it is judged that No.6 transformer has DC bias.