Research On Evaluation Method Of The DC Bias Effect Of GIC On The Traction Transformer

In recent years,China's high-speed railways have been developing rapidly and have become an important means of transportation for people to travel.The safety of high-speed railways is also increasingly drawing attention from all walks of life.when the extreme space weather geomagnetic storms occur,the disturbance of the geomagnetic field will induce ESP(Earth Surface Potentials,ESP).and when the ESP acts on long-distance transmission lines,it will generate GIC(Geomagnetically Induced Currents,GIC).The GIC can easily cause transformer accidents.The traction power supply system has many transformers.And traction transformers are the key equipments.They undertake the important task of safe transportation of high-speed railways.It is imperative to study the effects of GIC on the operating characteristics of traction transformers and assess their harmful effects.This thesis starts with the introduction of the structure of China's high-speed railway AT power supply system,and discusses the flow path of GIC in the high-speed railway AT power supply system and its approach to invading traction transformers.The fundamental reason of DC bias and excitation current distortion in traction transformers is analyzed in principle.And the first measured GIC data in the high-speed railway power supply system is given.Secondly,an equivalent model for the circulation loop of the GIC in the power supply system is established.Through the calculation and analysis,it is found that the degree of DC bias of the traction transformer caused by the GIC is related to the asymmetry of the equivalent resistance of the circulation loop.And based on that,this thesis improved the traditional traction network simulation model.By changing the model parameters,under the condition of different asymmetry degrees of equivalent resistance and different working conditions,the simulation of the DC bias of the traction transformer caused by the different amplitudes of GIC is completed.And then this thesis introduces the characteristics,classification and related applications of ANN(Artificial Neural Network,ANN).It is illustrated that the harmfulness of GIC to traction transformers can be evaluated using ANN by its good characteristic of nonlinear mapping.The thesis also introduces the basic structure of ANN-artificial neuron model,different learning methods and rules of ANN.Based on the basic principles of ANN,the model of BP neural network is introduced,which is used for GIC hazard assessment.The algorithm of BP neural network is deduced in detail,and the flowchart is given to realize the function of BP neural network.Finally,by comparing and analyzing the differences between various physical characteristics that reflect the degree of DC bias of traction transformers,this thesis selects the maximum value and distortion rate of traction transformer excitation current as the evaluation indexes.Using simulation data as the training sample,a BP neural network with double hidden layers is designed and trained,and it is also verified by the test sample.Network training parameters and evaluation results indicate that ANN can well evaluate the impacts of GIC on traction transformers,taking the excitation current maximum value and distortion rate as the evaluation indexes.