Research On Electromagnetic Vibration Characteristics And Looseness Identification Method Of Transformer Windings

With the continuous development of China’s economic level,the power industry,as a priority energy industry in economic development,has continuously expanded its scale and gradually moved toward ultra-high voltage,large capacity,and national interconnection.Transformer is an important equipment for power system voltage conversion,power distribution and transmission,and its safe and stable operation is essential.Due to the long-term operation of the transformer and the harsh working environment,environmental factors and power system failures will affect the operation of the transformer.Therefore,transformer failures occur frequently,especially the mechanical defects of the transformer are extremely common.Among them,the loosening of the windings is the main part.The “sub-health” state of transformer winding defect operation can easily develop into equipment failure or even cause electric power accident.Therefore,it is urgent to study the looseness of transformer winding and detect the mechanical state of transformer winding in real time,and timely arrange operation and maintenance inspection,which can effectively ensure transformer stability.In addition,it is of great significance to the safe and stable operation of the power system.In this thesis,firstly,This paper introduces the multi-degree-of-freedom vibration mechanics theory,and derives the analytic model of the transformer winding vibration based on the modal superposition method.It explains the relationship between the main mechanical parameters of the model and the vibration displacement response,and clarifies the mechanism of the pre-tightening force on the vibration characteristics of the winding.the transient electromagnetic field of the transformer is analyzed based on field-circuit coupling,and the electromagnetic force of the winding is calculated by the virtual displacement method.It is used as the excitation source of electromagnetic vibration,and the electromagnetic-mechanical coupling field is established to realize the harmonic response of the winding electromagnetic vibration under different pre-tightening forces.The simulation analysis results are consistent with the analytical model analysis.Finally,the design of the winding loosening experiment proves the validity of the analysis results.Based on the analysis of the vibration characteristics of the windings,this paper proposes a method for identifying the loose state of the windings.The method is based on adaptive complete integrated empirical mode decomposition(CEEMDAN)to decompose the vibration signal into multiple intrinsic modal function(IMF),extract the sensitive IMF component to calculate the sample entropy,and establish the winding feature vector set.Then,based on multi-class correlation vector machine(MRVM),the classifier is constructed,and the improved fruit fly optimization algorithm(IFOA)is used to optimize the MRVM kernel parameters,which greatly improves the accuracy of MRVM.Comparing this method with similar fault identification methods has obvious advantages in recognition speed and accuracy.The state recognition method in this paper can effectively identify the mechanical state of the winding,which is of great significance for the early fault detection of the winding.