Research On Fault Diagnosis Methods Of Transformer Windings Based On Spatial Vibration Distribution And Operational Modal Analysis

As a key device in the power system,large power transformers undertake the important tasks of voltage conversion and power distribution.Its long-term stable and reliable operation is of great significance for ensuring the power supply safety of the power system.Vibration analysis method is one of the important means of transformer condition monitoring and fault diagnosis.Compared with other types of transformer monitoring methods,it has no electrical connection with the inside of the transformer,has high sensitivity to the defects of the transformer mechanical structure,is safe and convenient,and is easy to not affect.The advantages of online detection under the condition of transformer operation have broad application prospects,and have been paid attention by scholars at home and abroad for a long time.Based on the summary of the previous research results of the transformer fault detection method based on the vibration analysis method,this paper starts from the perspective of extracting the fault characteristics of the transformer mechanical structure from the vibration signal,and analyzes the steady-state vibration and transient under the operating conditions of the transformer.Vibration feature extraction and winding mechanical fault diagnosis methods were discussed and researched in depth.By studying the vibration distribution characteristics of each measuring point at different positions on the surface of the transformer tank,this paper proposes a transformer winding fault diagnosis method based on spatial vibration distribution characteristics.This method can more sensitively and accurately reflect the mechanical structure state of the transformer winding.Realizing the monitoring and diagnosis of transformer winding faults provides a new and effective method and means.Based on the modal analysis method of the working condition,the transient vibration signal of the transformer during the power-off process is studied,and a new state evaluation and fault diagnosis method for the loosening of the transformer winding compression force is proposed by popular learning methods.The full text is summarized as follows:Firstly,introduces the background and significance of the subject,as well as the development history,research status and shortcomings of the transformer fault diagnosis algorithm based on the vibration analysis method.From the point of view of no previous involvement or insufficient research,it is proposed to make full use of the vibration distribution characteristics of multiple measurement points at different positions of the transformer tank wall,and the information contained in the transient vibration signal during the power outage of the transformer to perform feature extraction and The research goal of winding fault diagnosis clarifies the main research content,route and main innovations of the full text.Secondly,focuses on the distribution characteristics of the axial vibration of the transformer winding and the vibration of the tank wall under different mechanical conditions,and the effects of different load current levels on it.By artificially introducing faults including loosening of the pressing force,partial loosening of the winding,deformation of the winding,etc.on the winding,and measuring the axial vibration of the winding and the multi-point vibration of the transformer tank wall under the steady-state operating conditions excited by different load currents,to study the current winding The change characteristics of the spatial vibration distribution when the mechanical structure failure occurs provides a basis for extracting the feature quantity that can reflect the health status of the winding mechanical structure from the vibration distribution.The experimental results show that the spatial vibration distribution characteristics are closely related to the health of the winding mechanical structure.When the winding mechanical structure changes,the morphological characteristics of the spatial vibration distribution,symmetry,and changes with load current have significant changes,indicating that There is indeed information in the spatial vibration distribution that reflects the mechanical structure of the winding.Thirdly,based on the experimental results obtained in Chapter 2,the variation trend of the transformer spatial vibration distribution characteristics under various fault types,combined with the vibration transmission law in the power transformer,the study proposed a feature extraction method based on the spatial vibration distribution of the transformer tank wall.Calculate the four quantitative parameters of spatial vibration distribution characteristics to quantitatively describe the spatial difference between the vibration signals at different measurement points on the transformer tank wall,and then reflect the state of the winding mechanical structure.On this basis,the above-mentioned quantized parameters of spatial vibration distribution characteristics are combined into feature vectors,and a transformer winding fault diagnosis model is established based on support vector machines to realize the analysis and diagnosis of the transformer winding state and faults.The field test data of the actual transformer shows that the diagnosis model can accurately and effectively realize the fault diagnosis of the transformer winding mechanical structure.Fourthly,the transient vibration characteristics of the winding during the power failure of the transformer,and the effect of loose winding compression on the transient vibration characteristics are studied.Based on the classical winding vibration mechanical model,with theoretical analysis and numerical analysis-based simulation analysis as the means,the main factors affecting the distribution of the winding transient vibration signal energy in the frequency domain and the transient The frequency domain characteristic of the state vibration signal changes with the loosening of the winding pressing force.Through the collection and analysis of the transient vibration signal of the winding of an experimental transformer during power-off,the conclusions of theoretical analysis and numerical simulation are verified,and the frequency domain characteristics of vibration are extracted with higher frequency resolution as The basis of fault detection and condition evaluation of winding loosening pressure based on transient vibration during power failure.Fifthly,on the basis of the frequency domain characteristics of the transient vibration signal of the aforementioned transformer power-off process,a manifold learning dimension reduction algorithm Limited Looseness Geodesic Gaussian Locality Preserving Projections(LLGGLPP)algorithm is proposed and applied It is used to diagnose loosening of winding pressure based on transient vibration signals.The algorithm is based on the LPP algorithm,combined with the changing characteristics of the transient vibration characteristics in the loose winding compression state,using approximate geodesic distance calculation,finite relaxation adjacency relationship construction and generalized Gaussian function to adjust sample spacing and other methods.The low-dimensional space with reduced dimensions better describes the manifold structure where the transient vibration characteristics gradually change as the winding compression force decreases.In order to give a quantitative estimate of the compression force of the winding,this paper measured the stress-strain characteristic curve of the insulating pad material of the experimental transformer,and combined with the obtained description of the low-dimensional spatial manifold structure,a transformer-based The winding compression force estimation method of transient vibration signal in electrical process provides an effective reference and means for further online monitoring and diagnosis of power transformer winding looseness.Finally,the main research contents of this paper are summarized,and the further research and improvement direction are prospected.