Identification Method Of Fault Area And Fault Type Of Distribution Network Based On Feeder Terminal Unit

The power industry is a pillar industry that supports the sustained and stable development of the national economy,and plays an irreplaceable role in all aspects of modern social development.Therefore,the reliable supply of electricity is an important guarantee to maintain the stability of the development.As the important infrastructure for guaranteeing the People's livelihood and the last key point of servicing to residents,the stable and safe operation of distribution network is directly related to the electricity interests,electricity quality and electricity reliability of customers.With the voltage level increasing and the integration of distributed generation,distribution network becomes increasingly complex.When faults occurs in distribution network,in order to prevent the expansion of the fault,it's urgent to locate the fault area,and overhaul to restore power supply.Therefore,the rapid and accurate fault processing of distribution network,including fault location and fault type identification,is the essential for the safe and stable operation of distribution network.Considering that the traditional protection of distribution network exists complicated relay settings,poor coordination and malfunction,the main research of this paper is organized as follows:Firstly,the theory of wavelet analysis and support vector machine classification are introduced.It is proposed that the fault feature(wavelet entropy)is extracted by the wavelet analysis,and the fault feature data is classified by the support vector machine(SVM)method after the fault occurring.By introducing the example analysis?the accuracy and applicability of the support vector classification method are validated.Secondly,according to the shortcomings of the existing power distribution network protection method,the fault location algorithm of distribution network based on MDS and LOF algorithm is put forward.This algorithm makes full use of the real-time data uploading by the wide area measurement system,and uses zero sequence,negative sequence,positive sequence three-phase current and zero sequence active,reactive power and other electrical quantity as the fault feature quantities.The basic idea of this method is to utilize the data uploaded by each sensor,to form a multidimensional state monitoring matrix through data preprocessing and data fusion.According to the MDS analyses of state monitoring matrix and the LOF value of each node,the proposed protection method can detect and locate the fault occurring in distribution network within a short time.The proposed multidimensional scale analysis and local outlier factor detection algorithm avoid the complex calculation of relay setting,and it could detect and locate fault even if the sensor devices is out of normal operation.Finally,considering the difficult of fault type identification after fault location,this paper proposes a novel fault identification method based on based on wavelet analysis and support vector.First.the wavelet signal is decomposed by wavelet analysis theory.And then,combined with singular value decomposition and information entropy theory,the voltage signal is processed to reflect the wavelet singular entropy of the fault feature,and the fault phase can be discriminated according to its size.Finally,he fault is classified into phase fault and ground fault by using the first-order low-frequency energy of the zero-sequence voltage,which reflects the ground fault information.The SVM classification model is used to verify the accuracy and validity of the proposed algorithm.