• DocumentCode
    1758574
  • Title

    Directional relays without voltage sensors for distribution networks

  • Author

    Trung Dung Le ; Petit, Marc

  • Author_Institution
    Dept. of Power & Energy Syst., E3S SUPELEC Syst. Sci., Gif-sur-Yvette, France
  • Volume
    8
  • Issue
    12
  • fYear
    2014
  • fDate
    12 2014
  • Firstpage
    2074
  • Lastpage
    2082
  • Abstract
    In this study, algorithms for directional relays using only current measurements are presented. Developed for radial distribution networks, these algorithms will determine fault direction based on ratios between variations of sequence currents during and before faults: ΔI2I0 ratio for line-to-ground faults and the ΔI2I1 ratio for line-to-line faults. The ratios are used as input of a support vector machine classifier, which was trained beforehand thanks to simulation tools. The classifier classifies ratios into two categories, according to the fault location: upstream or downstream towards the relay. Test results from simulations show good performances of the algorithms in most cases, with the presence of different distributed generation technologies. Moreover, impact of certain factors on algorithms, such as measurement errors, high-impedance faults or network reconfiguration, is studied. Finally, the implementation of algorithms is also discussed.
  • Keywords
    distributed power generation; electric current measurement; fault location; power distribution faults; power distribution protection; relay protection; support vector machines; current measurements; directional relays; distributed generation; fault direction; fault location; high-impedance faults; line-to-ground faults; line-to-line faults; measurement errors; network reconfiguration; radial distribution networks; support vector machine classifier; voltage sensors;
  • fLanguage
    English
  • Journal_Title
    Generation, Transmission & Distribution, IET
  • Publisher
    iet
  • ISSN
    1751-8687
  • Type

    jour

  • DOI
    10.1049/iet-gtd.2014.0308
  • Filename
    6985807