Exploration versus exploitation using kriging surrogate modelling in electromagnetic design

• Song Xiao ^[1] ; Mihai Rotaru ^[1] ; Jan K. Sykulski ^[1]
  1. [1] University of Southampton

     University of Southampton

     GB.ENG.M4.24UJ, Reino Unido

     
• Localización: Compel: International journal for computation and mathematics in electrical and electronic engineering, ISSN 0332-1649, Vol. 31, Nº 5 (Special Issue: Electromagnetic Fields in Electrical Engineering), 2012, págs. 1541-1551
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Purpose – Design optimisation of electromagnetic devices is computationally expensive as use of finite element or similar codes is normally required. Thus, one of the objectives is to have efficient algorithms minimising the number of necessary function calls. In such algorithms a balance between exploration and exploitation needs to be found not to miss the global optimum but at the same time to make efficient use of information already found. The purpose of this paper is a contribute to the search of such efficient algorithms.

    Design/methodology/approach – This paper discusses the use of kriging surrogate modelling in multiobjective design optimisation in electromagnetics. The investigation relies on the use of special test functions.

    Findings – The importance of achieving appropriate balance between exploration and exploitation is emphasised when searching for the global optimum. New strategies are proposed using kriging.

    Originality/value – It is argued that the proposed approach will yield a procedure to solve time consuming electromagnetic design problems efficiently and will also assist the decision making process to achieve a robust design of practical devices considering tolerances and uncertainties.