• DocumentCode
    1541854
  • Title

    A combined technique for the global optimization of the inverse electromagnetic problem solution

  • Author

    Borghi, C.A. ; Fabbri, M.

  • Author_Institution
    Dept. of Electr. Eng., Bologna Univ., Italy
  • Volume
    33
  • Issue
    2
  • fYear
    1997
  • fDate
    3/1/1997 12:00:00 AM
  • Firstpage
    1947
  • Lastpage
    1950
  • Abstract
    A method for finding the global optimum of the electromagnetic inverse problem solution is presented. The problem consists in the determination of the current system which produces a given magnetic field within a domain of finite extent. An infinite class of solutions exists. In order to find the solution of the synthesis problem obtained, a global optimization technique, based on the combination of the filled function (FF) method and the evolution strategy (ES) method, is developed. The FF method is utilized to find a sequence of regions containing decreasing minima. The minimization inside this region is done by means of the ES method. The technique described has been utilized for the poloidal magnetic system of a toroidal device for thermonuclear fusion experiments and for the main coil system of a nuclear magnetic resonance (NMR) device
  • Keywords
    coils; electromagnetism; fusion reactors; inverse problems; magnetic fields; nuclear magnetic resonance; optimisation; plasma toroidal confinement; current system; evolution strategy method; filled function method; finite extent domain; global optimization; inverse electromagnetic problem solution; magnetic field; main coil system; minimization; nuclear magnetic resonance device; poloidal magnetic system; synthesis problem solution; thermonuclear fusion experiments; toroidal device; Cost function; Current density; Evolution (biology); Inverse problems; Magnetic devices; Magnetic resonance; Minimization methods; Nuclear magnetic resonance; Optimization methods; Toroidal magnetic fields;
  • fLanguage
    English
  • Journal_Title
    Magnetics, IEEE Transactions on
  • Publisher
    ieee
  • ISSN
    0018-9464
  • Type

    jour

  • DOI
    10.1109/20.582674
  • Filename
    582674