Research On Fault Diagnosis Method Of Power Transformer Winding Based On Vibration Analysis Method

With the continuous development of modern electric power industry,higher requirements are put forward for improving the safety and stability of electric power equipment.Carrying out status monitoring and status diagnosis of power transformers can obtain real-time internal mechanical status of power transformers,and can prevent accidents in time,which is of great significance to ensure the safety and stable operation of my country’s power system.In order to solve the problems of data acquisition difficulties and offline diagnosis existing in the current transformer winding state evaluation.Carry out research on the application of vibration analysis method in power transformer winding fault diagnosis technology.Through in-depth research on power transformer vibration mechanism,winding mechanical state simulation,vibration signal analysis,signal feature extraction,fault identification algorithm,etc.,the application of power transformer fault diagnosis in actual engineering is realized.The details of the work are described below:(1)Analysis of mechanical vibration mechanism of power transformers.Study the vibration characteristics of transformer windings,the mathematical model of winding vibration,and the transmission of transformer vibration energy,and conduct in-depth analysis of vibration principles.Based on actual product parameters,a vibration simulation model of a40000 k VA/110 k V power transformer is established.Based on the analysis of the transient structure field,the weak position of the transformer winding is judged,and the winding deformation fault state samples are provided for the research of transformer fault identification.(2)Research on mechanical vibration characteristics of power transformer windings.The electric-magnetic-solid multi-physics coupling simulation calculation is performed on the vibration simulation model,and the winding deformation trend obtained by the simulation and the vibration characteristics of different measuring points on the surface of the fuel tank are analyzed.A load experiment of 40000 k VA/110 k V power transformer was carried out,and the accuracy of the model was verified by comparing with the experimental results.On the basis of the transformer model under normal operating conditions,the transformer fault simulation models were designed for the transformer winding radial deformation,the loosening of the pretightening force,and the inter-turn short-circuit fault state.The simulation calculation analyzes the vibration characteristics of the winding under different fault conditions.The vibration characteristics under different load current conditions and at each measuring point position are studied.(3)Extraction of eigenvalues of power transformer winding vibration.The analysis method of winding vibration signal based on wavelet packet decomposition is proposed.By decomposing the sub-bands of the signal,the corresponding relationship between frequency band and energy entropy is established as the characteristic vector of the vibration signal.The difference between the characteristic vectors of the winding vibration signal under different mechanical conditions is used to lay the foundation for the diagnosis of the transformer condition.(4)Development of diagnostic software for power transformer winding status.The neural network model of the extreme learning machine is applied to the winding fault diagnosis.For the sensitive problem of parameter selection,particle swarm algorithm is used to optimize the kernel function of extreme learning machine,and a diagnostic program for power transformer winding state is developed.The results of example analysis show that the proposed method can effectively diagnose the winding state of the transformer,and has a better performance in efficiency and accuracy.