Study On Fault Diagnosis Of Traction Transformer Winding And Core Based On Vibration Signal

The traction transformers are important in the electrified railway system,safety and reliability,so if we can find the potential security risks in the traction transformer in time,not only can improve transportation efficiency,but also reduce maintenance cost.In the existing methods of transformer monitoring,the vibration analysis method is sensitive to the internal mechanical structure failure,and the vibration method has no electrical connection with the traction transformer,which is a very effective method for the electric monitoring of the traction transformer.Based on the method of vibration signal analysis,the core and winding of the traction transformer are used as the research object,and the characteristic information of the core and winding respectively in normal state and fault state is studied by means of theoretical analysis and experiment.In this work,firstly,from that vibration mechanism of traction transformer,the mathematical model of winding vibration is established by the spring-mass model,and the relationship between the size of the winding force and the vibration acceleration is obtained.At the same time,according to the theory of magnetic mechanics,the main factors that cause the core vibration are discussed,and the relationship between the core vibration signal and the core condition is discussed.And combining with the results of transformer fault detection at home and abroad,adopts the method of extraction of vibration signal characteristic value,as the basis of the vibration fault traction transformer is given,which based on the change of characteristic values in different condition determine the core and winding faults.Then a set of vibration data acquisition system is designed.Its main function is to acquire that vibration signal of traction transformer,transmit the acquired signal to the upper computer,and the upper machine signal processing use MATLAB to judge the fault message.Finally,the design of the vibration data acquisition system is used for field test.The test results show that the vibration signal acquisition system has good performance and can be applied in practice.Signal eigenvalue extraction is the difficulty of fault diagnosis as the core of this article,compared the advantages and disadvantages of various signal processing methods in recent years,and considering the traction transformer field environment is complex,low signal noise ratio(SNR)acquisition of vibration signal,this paper puts forward the combination of VMD and sparse decomposition algorithm to extract the signal characteristic information,and through the simulation signal processing,found that the algorithm can extract characteristic frequency in strong noise environment.Apply the algorithm to the collection of vibration signal of the core and winding,the winding is normal and fault condition signals,obtained the winding fault threshold,and the threshold can be used as the basis for judging winding fault.Core vibration is relatively complex,through the state of normal and fault vibration signal processing core,core vibrational eigenvectors are extracted,for in the face of core fault intelligent diagnosis after laid a solid foundation.Finally,the paper applies BP neural network and the algorithm of PNN network core fault intelligent diagnosis,through the input feature vector to determine the core fault,at the same time prove that PNN algorithm than the BP neural network algorithm has better advantages in the core of fault diagnosis.