Research On Transformer Fault Diagnosis Based On Vibration Analysis Method

With the rapid development of my country’s power market economy in recent years,the capacity of power grids is constantly increasing.The power industry is the most important basic energy industry in the development of the national economy,and is the priority development focus and basic industry in the economic development strategies of various countries.The safety of power transformers is an important guarantee for the safe,reliable,quality and economic operation of the power grid system.Traditional methods are based on the impedance,capacitance,inductance,mutual inductance and changes in the gas generated by insulation aging to monitor the state of the transformer.The vibration analysis method is not only fast when analyzing faults,but also has the huge advantage of not directly affecting the actual normal working system operation of the entire power transformer because there is no electrical connection compared to other methods.This article starts with the common faults and fault classification of transformers,and introduces the traditional methods of detecting transformer windings and cores.Compared with the traditional methods,the vibration analysis method is proposed.The mathematical model of the winding is established to further explain the force of the winding under the influence of the leakage magnetic field and the short-circuit current,and the relationship between the winding vibration acceleration amplitude,the load current and the frequency;the magnetostriction phenomenon of the silicon steel sheet in the magnetic field is studied to illustrate Correlation of core vibration acceleration amplitude,power supply voltage and frequency.Using a combination of theoretical analysis and actual measurement verification,the vibration signal characteristics of windings and cores during normal operation and faulty operation are studied,and the fault monitoring of windings and cores of power transformers is carried out by vibration analysis.The main content of the research is the classification of power transformer faults and its vibration mechanism,and then a vibration fault monitoring platform is established,which includes the selection and installation of acceleration sensors,circuit design,and finally the vibration data of the transformer during normal operation.In order to solve the expensive problems of acceleration sensors,operational amplifiers,and industrial computers,an embedded data acquisition system based on STM32 was designed.Through the application of single-chip microcomputer STM32F103c8t6,AD7606 module,using FreeRTOS operating system to collect.Algorithmically,the vibration data is analyzed through de-envelope,and it is demonstrated that the existence of transformer faults and the rationality of the fault types can be obtained through the vibration data.Finally,perform Hilbert-Huang algorithm analysis on the normal data of the transformer actually collected,apply empirical mode decomposition to obtain the eigenmode function and residual quantity,and use Hilbert for each eigenmode function obtained by decomposition.Transform to obtain a three-dimensional time-spectrogram of time,frequency,and amplitude to predict the potential risk of transformer failure.Finally,the final envelope spectrum of the Hilbert-Huang transform is obtained.For this large transformer,when the transformer is operating in a steady state,it reaches the peak frequency at about 50 Hz.When the transformer in normal operation is at the moment of power-on,it reaches the peak frequency at about 60 Hz.From this,it can be qualitatively analyzed that the state of the transformer is related to the state of the transformer when the hertz reaches the peak value.Perform algorithmic analysis on each large transformer to obtain daily data,and then when there is a large difference between the daily frequency value,it is inferred that it has a risk of failure or has failed.