Research On Line Selection Method For Single-phase Grounding Fault In Small Current Grounding System

The small current grounding system has high reliability of power supply,and the system can operate normally for a short time after a single-phase grounding fault occurs.However,if the faulty line cannot be removed in time,It will lead to the sound line insulation is destroyed or even upgraded to a multi-phase fault,which will destroy the safe and stable operation of the power system.Therefore,it is of great significance to study a reliable and effective line selection method.However,due to the inconspicuous characteristics of the steady-state fault and the short transient fault signal,the accuracy of the existing line selection methods is unsatisfactory.Based on this,this paper studies a support based on Ensemble Empirical Mode Decomposition（EEMD）,Hilbert marginal spectrum,energy entropy,and optimized by genetic algorithm（GA）.A vector machine（support vector machine,SVM for short）is employed to make the method of fault line selection come true.In this thesis,after the fault,the transient and steady state fault features of the small current ground system are analyzed theoretically.Therefore,the transient zero-sequence current is chosen for line selection research and a simulation model is built in Simulink.The simulation waveform is compared with the theoretical analysis,so the validity of the model is confirmed,so as to provide reliable data for subsequent line selection research.Secondly,the transient zero-sequence current is preprocessed by mathematical methods such as EEMD decomposition and Pearson correlation coefficient.After the zero-sequence characteristic current is extracted,the energy entropy of the Hilbert marginal spectrum of the zero-sequence characteristic current of each line is calculated to form the characteristic vector.Finally,the fault feature vector is input into the SVM classifier to determine the faulty line.However,since the two parameters when the SVM uses the RBF kernel function are only set by experience,the accuracy of the line selection will be influenced.This thesis introduces the genetic algorithm to optimize SVM parameters,in order to make line selection more accurate.In order to verify the effectiveness and accuracy of the line method proposed in this paper,a variety of faults are simulated and calculated in the simulation model,and multiple sets of fault feature vectors are obtained,which are respectively input into SVM and GA-SVM for training and testing.The superiority of GA-SVM and the validity and accuracy of the line selection method raised in this thesis are verified by comparing the classification resultsAnd the fault feature extraction is carried out for the transient zero-sequence current under different grounding mediums in the single-phase grounding fault test,and the universality of the line method proposed in this paper is proved by comparing the energy entropy value.