Study On The Power Transformer Winding Vibration Simulation And State Detection

The power transformer is the main electrical equipment in power grid.The growing scale and increasing short-circuit capacity of power grid have made power transformer prone to looseness as well as deformation fault danger under sudden short-circuit impact,which causes great threat to the safe and stable operation of power grid.Therefore,it is of great significance for power transformer and the whole power system to study on the accurate detection method of winding state.The power transformer winding condition detection method based on vibration technology has been a hotspot of research for its high sensitivity and good anti-interference ability.To further improve its detection ability,this paper focuses on the winding vibration mechanism,studies sweep-frequency experiment and fault diagnosis of power transformer.The main works are as followsTo analyze the vibration response of transformer winding accurately,this paper uses finite element simulation based on actual geometric parameters,and calculates magnetic field,electro-dynamic force and vibration response with a thought of multi-field coupled simulation.The results are in good agreement with experiment signal,verifies the accuracy of modeling and simulation.Besides,the results also show the distribution disciplinarian of magnetic field and electro-dynamic force.The change of compression force will increase the vibration amplitude and harmonic component,and the higher frequency has greater increase.According to the vibration frequency response principle,this paper designs and builds the sweep-frequency experiment system for large power transformer.In the experiment,vibration signals as well as calculated vibration frequency response curves under normal and typical fault conditions are obtained by setting fault for some 110 k V transformer.This paper calculates vibration severity,gray relativity and correlation coefficient for different vibration frequency response curves,which verifies good repeatability of this method.The results also show that vibration frequency response curve is closely related to the transformer winding state,the chosen indicators have certain detection ability for winding faults.In order to realize effective fault diagnosis for transformer winding,the singular value decomposition of bispectrum via auto-regressive moving average model and improved fuzzy clustering algorithm method is proposed by this paper,according to the nonlinearity of vibration signal.Firstly,the bispectrum is calculated by using ARMA model analysis and the characteristic matrix is constructed according to cross slices of bispectrum.Then,singular value decomposition is applied to the matrix for dimensionality reduction,which gives the fault feature vector.Finally,the improved fuzzy clustering algorithm is used to classify these singular value features.Thus realizing accurate identification of different fault types of transformer winding.Results show that this method could effectively diagnose looseness as well as multiple deformation faults of transformer winding.The findings of this paper can provide theoretical basis and reference for the engineering application of transformer winding vibration detection technology.