Simulation And Identification Of Winding Deformation Of Oil-immersed Power Transformers

With the rapid development of China's economy,it is more and more important to build a safe and reliable power grid.It has become a primary goal to ensure the safe and stable operation of power grid equipment.The role of oil-immersed power transformers in the power grid is irreplaceable.It has been found that the deformation of the internal windings is one of the main causes of transformer failure.Based on this,it is of great significance to explore the simulation and identification of oil-immersed power transformer winding deformation.In this paper,we first compare the detection methods of transformer winding deformation commonly used at home and abroad,and finally choose the frequency response analysis method,and develop a solution for its detection shortcomings.Secondly,comprehensive analysis and research on the identification technology of electric transformer winding deformation proposed by domestic and foreign scholars,and considering the frequency response analysis curve to detect the deformation of the transformer winding and the obtained experimental sample data,the manual identification of the experimenter is high and difficult.It is proposed to use the extreme learning machine to identify the type of winding deformation.Based on the selection of frequency response analysis method as the detection method of winding deformation,the transformer winding equivalent circuit is established.For the calculation of electrical parameters in the equivalent circuit model,two methods are used in this paper: using analytical formula method and finite element method,and comparing them.Finally,through analysis and comparison,it is concluded that the finite element method is more accurate in calculating electrical parameters.Based on the parameters obtained by the finite element method,the PSpice software is used to establish the winding equivalent circuit model,and the frequency response characteristic curves of the normal winding and the windings of different fault types are obtained.The curve is a wave and a trough,and the peaks and troughs alternate Oscillation curve.Compare the frequency and amplitude of the resonance point on the normal and fault winding frequency response curves to detect the winding deformation.There is an upper limit frequency for the lumped parameter circuit model.If the frequency is close to the upper limit frequency,the internal series capacitance will change the distribution of the winding voltage.Generally,the winding propagation characteristics can be fully characterized in 1MHz.This paper proposes to select the swept frequency signal for the injected signal source from 1kHz to 1MHz.The square wave pulse is selected when detecting the fine deformation of the winding in the high frequency band.Finally,the analysis of transformer winding deformation fault detection results mainly relies on manual judgment and analysis.Experts and scholars at home and abroad have also proposed algorithms based on neural network,support vector machine to realize deformation type identification,and eliminate human error,but also there are some shortcomings.In this paper,a method based on nuclear limit learning machine for transformer winding deformation identification is proposed.At the same time,it is very sensitive to its parameter setting.In this paper,genetic algorithm is used to optimize the parameters of nuclear limit learning machine,and then the optimization parameters are applied to the nuclear limit learning machine.In the algorithm,the kernel limit learning machine is finally applied to complete the deformation recognition.