Open AccessFault Line Selection Method for Distribution Network Based on Angle Similarity
Author(s) -
Guang Feng,
Ming Chen,
Guojie Xiang,
Kun Yu,
Xiangjun Zeng
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/495/1/012024
Subject(s) - similarity (geometry) , cluster analysis , fault (geology) , line (geometry) , feature (linguistics) , data mining , pattern recognition (psychology) , sample (material) , selection (genetic algorithm) , feature selection , computer science , artificial intelligence , fuzzy clustering , fuzzy logic , mathematics , image (mathematics) , physics , linguistics , philosophy , geometry , seismology , thermodynamics , geology
Aiming at the problem that the current distribution network protection technology has a low accuracy of line selection, the method of fault line selection based on angle similarity is proposed. A plurality of fault feature quantities of each feeder in different operating states are collected to form a historical database. The clustering centers of fault classes and non-fault classes are obtained by fuzzy clustering analysis. Similarly, the real-time data of each feeder is collected to form a sample of the feature to be tested. The degree of similarity between the feature samples to be tested and the historical feature samples is analyzed by angle similarity. The sample to be tested is classified into a fault class or a non-fault class, and finally the fault feeder is accurately selected.