Research On Fault Line Selection Of Small Current Grounding System Based On Fault Transient Feature Clustering

According to the data of distribution network operation,single-phase ground fault occurs most frequently,accounting for more than 70% of all fault types.Aiming at the problem of line selection of single-phase-to-ground fault,because the fault situation is complex,and the fault current is affected by neutral grounding mode,noise and other factors,the accuracy of line selection has been low.When a single-phase ground fault occurs,if the fault line can not be selected quickly and accurately,it is very likely to develop into a more serious phase-to-phase short circuit,which will eventually lead to equipment damage and affect the power supply reliability of the system.Firstly,this article systematically introduces the advantages and disadvantages of the existing fault line selection methods and the scope of application,and analyzes in detail the characteristics of the zero sequence current of each line after the failure of the small current grounding system.According to the difference between the transient zero sequence current of the faulted line and the non-faulty line,this article studies a fault line selection algorithm based on correlation clustering.This method does not require a large number of complicated calculations,which greatly improves the efficiency of fault line selection.The zero-sequence current signal is often affected by the neutral grounding method,noise and other various factors.In order to perform fault line selection more reliably,based on the fault line selection algorithm of correlation clustering,a fault line selection method based on clustering of non-power frequency components is proposed to solve the problem that directly using correlation clustering algorithm cause the inaccuracy of line selection.First,the variational modal decomposition and fast Fourier transform are combined to accurately separate the power frequency components and noise signals in the transient zero sequence current to obtain the non-power frequency components.The Pearson correlation coefficient is used to calculate the correlation coefficient matrix of non-power frequency components of each line,then uses the correlation coefficient matrix as the input of K-means clustering,and finally achieve the purpose of precise line selection through the correlation clustering algorithm.This method not only takes into account the efficiency of the correlation clustering line selection algorithm,but also improves the accuracy of the fault line selection.Meanwhile,it can also avoid the inaccurate separation of the power frequency components and the correlation selection caused by the traditional algorithm due to factors such as modal aliasing,and it can avoid the problem of threshold selection in the method of correlation line selection.The simulation results show that the line selection method proposed in this paper can accurately select the faulty line under various conditions.The result of line selection is not affected by factors such as fault resistance and fault initial phase angle,and it also has strong anti-interference ability.