• DocumentCode
    2485682
  • Title

    Modeling of a high voltage insulator under uniform and non uniform pollution

  • Author

    Bencherif, Y.A. ; Mekhaldi, A. ; Teguar, M.

  • Author_Institution
    Lab. de Rech. en Electrotech., Ecole Nat. Polytech. d´´Alger, Algiers, Algeria
  • fYear
    2012
  • fDate
    14-17 Oct. 2012
  • Firstpage
    761
  • Lastpage
    766
  • Abstract
    Our study presents a modeling of a high voltage polluted insulator. This model is used for the calculation of the flashover voltage under pollution conditions. Two methods are proposed in this paper. In the first one, we propose a dynamic model predicting the propagation of the discharge on the polluted surface. We use an equivalent electrical circuit and the propagation criteria of Hampton. The shape of the insulator is considered by using its open model. This latter is a transformation of the insulator´s profile to a 2D representation. The validation of the calculation results is done using experimental studies. In the second one, a regression model of the insulator is proposed using the Least Square Vector Machine (LS-SVM) method. A Hybrid optimization technique is proposed in order to improve the prediction performances of this supervised learning statistical method. Tests and validations are done in order to evaluate the viability of the proposed model. The ability of prediction of the two presented models is tested. According to the applied investigations, it was noted that the optimized LS-SVM model offers better performances.
  • Keywords
    flashover; insulator contamination; learning (artificial intelligence); least squares approximations; optimisation; power engineering computing; regression analysis; support vector machines; 2D representation; LS-SVM model; dynamic model; equivalent electrical circuit; flashover voltage; high voltage polluted insulator modeling; hybrid optimization technique; insulator profile; least square vector machine method; nonuniform pollution; pollution conditions; prediction performances improvement; propagation criteria of Hampton; regression model; supervised learning statistical method; uniform pollution; Conductivity; Flashover; Insulators; Integrated circuit modeling; Kernel; Mathematical model; Pollution;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Electrical Insulation and Dielectric Phenomena (CEIDP), 2012 Annual Report Conference on
  • Conference_Location
    Montreal, QC
  • ISSN
    0084-9162
  • Print_ISBN
    978-1-4673-1253-0
  • Electronic_ISBN
    0084-9162
  • Type

    conf

  • DOI
    10.1109/CEIDP.2012.6378892
  • Filename
    6378892