Simulation Study On Monitoring And Recognition Power Transformer Winding Deformation Based On Finite Element Method

Power transformer is one of the most important electrical equipment in power system. Its health has great significance to the whole safe operation of the power grid. Windings fault is the one of the most common transformer faults, which cannot be easily found. The deformation has accumulation effect, when the winding deformation becomes violent, the occurrence of the severe external short circuit will lead to serious deformation of windings, which will destroy the transformer. So exploring the reasons of power transformer winding deformation, monitoring and recognition the winding deformation are very important and necessary.In this paper, the distribution characteristics of electromagnetic force on windings when three-phase short-circuit of transformer second winding happened is got by finite element method, it can predict the types of winding deformation; Equivalent circuit parameters of healthy and defamation winding are also calculated by finite element method; Based on the circuit parameters, the radial displacement fault, radial deformation and space between disks change are monitored by frequency response analysis; The recognition of winding deformation types by SVM is realized through data got by frequency response analysis signature. The major achievements are as follows:① The distribution characteristics of electromagnetic forces is got by 2-D finite element model of power transformer. The 2-D model of transformer is built in finite element software, and it is verified by analytical method; Types of winding deformation can be predicted though the distribution characteristics of short-circuit force, which is the basic of monitoring and identifying the situation of winding deformation.② The computing method of parameters of equivalent circuit of healthy and deformation winding in power transformer is proposed. The turn-to-turn capacitance, disk-to-disk capacitance, ground capacitance and inductance are calculated based on analytical method and finite element method; The results got by the two methods are compared, which shows the finite element method is more accurate, and it also can calculate the circuit parameter of deformed winding.③ The regularity results of the change of equivalent circuit parameters and frequency response analysis signature caused by winding deformation are indicated. The models of winding with different deformation types and degrees have been established in finite element software ANSYS Maxwell, and the winding deformation are simulated by the capacitance and inductance got by the models; The winding equivalent circuit is simulated in PSPICE based on the circuit parameters to get the frequency response analysis signature of deformed winding; Summarize the regularity results of the change of equivalent circuit parameters and frequency response analysis signature when winding deformation happen, in order to monitor winding deformation.④ Recognition of power transformer winding deformation types is realized by support vector machine(SVM). The variation of frequency and amplitude of resonance point and correlation coefficient of low, middle and high frequency band are got from frequency response analysis signature of windings, they are as characteristic quantity in train and test sample to recognize winding deformation types; The recognition results show the variation of frequency and amplitude of resonance point as characteristic quantity is better in recognition result by SVM, which has ideal recognition result.