Research On Transformer Online Monitoring Based On Tikhonov Regularization Method

The transformer is one of the important equipment in the power system.During the installation and operation of the transformer winding,under the action of mechanical external force and electromotive force,deformation may occur.When the winding deformation accumulates to a certain extent,it will cause damage to the main insulation of the transformer,which will lead to more serious faults such as a short circuit between turns.At present,the detection methods of transformer winding deformation are mostly offline detection,but the shortcoming of offline detection method is that it needs to be detected after the transformer is powered off.Therefore,it is very necessary to explore the method of online monitoring transformer winding deformation.This paper analyzes the existing transformer equivalent model,selects the T-type equivalent model as the research object,and establishes the transformer short-circuit impedance calculation equation,and judges the transformer winding state by online monitoring the transformer winding short-circuit impedance parameters.However,when solving the equations of winding parameters,due to the ill-conditioning of the calculation equations,when the commonly used least square method is used for calculation,small errors in the measured values will cause large changes in the calculation results and affect the accuracy of the short-circuit impedance parameter calculation.To solve this problem,this paper introduces the Tikhonov regularization calculation method,and proposes a method for online transformer monitoring winding deformation based on Tikhonov regularization to improve the calculation accuracy.After determining the calculation method,this article uses electromagnetic transient simulation to perform simulation calculations.In the calculation,the Tikhonov regularization algorithm is compared with the least square method.The study found that the commonly used least square method is not accurate enough for the short-circuit impedance parameter resistance component calculation.The calculation error is up to 112.54%,and the calculation error of the Tikhonov regularization algorithm for the transformer short-circuit impedance resistance parameter and reactance parameter is less than 1%.The calculation result is not affected by the load rate,power factor and short-circuit fault,and the calculation data When the window length is a half cycle of the power frequency,the short-circuit impedance parameters of the transformer can still be accurately calculated,and the rapid identification of the short-circuit impedance parameters of the transformer can be realized.On the whole,the computational performance of Tikhonov’s regularization algorithm is better than that of the least square method.Finally,under the existing conditions of the laboratory,an equivalent transformer model and a toroidal transformer experimental platform were built to verify the Tikhonov regularization algorithm.The experimental results show that the error of the Tikhonov regularization method to identify the short-circuit impedance parameters of the transformer is less than 1%,which is the experimental result.Consistent with the simulation results,it is further confirmed that the Tikhonov regularization algorithm can realize the application of on-line monitoring of transformer winding deformation.