Comparison of multi‐objective optimisation approaches for inverse magnetostatic problems

• S. Carcangiu ^[1] ; P. Di Barba ^[2] ; A. Fanni ^[1] ; M.E. Mognaschi ^[2] ; A. Montisci ^[1]
  1. [1] University of Cagliari

     University of Cagliari

     Cagliari, Italia

     
  2. [2] University of Pavia

     University of Pavia

     Pavía, Italia

     
• Localización: Compel: International journal for computation and mathematics in electrical and electronic engineering, ISSN 0332-1649, Vol. 26, Nº 2 (Selected papers from the 9th Workshop on Optimization and Inverse Problems in Electromagnetism, Sorr), 2007, págs. 293-305
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Purpose – The aim of the paper is to compare two different approaches to multi‐objective optimisation in magnetostatics; in this way, the case study is investigated as a possible benchmark.

    Design/methodology/approach – A Tabu search method modified with ε‐constraint algorithm is compared with a multi‐objective multi‐individual evolution strategy. The case study is the automated shape design of a magnetic pole. In order to reduce the computational cost of solving the direct problem, which requires repeated analyses of the magnetic field, a neural network has been used to approximate the objective functions that depend on the design variables.

    Findings – An approximation of the Pareto front for each method is obtained. A twofold comparison between the two methods is made, based on both the result accuracy and the computational cost.

    Originality/value – Two different methods were already tested on a case study proposed as a benchmark for multi‐objective optimization in magnetostatics. The paper represents a contribution to bridge the gap between analytical and numerical benchmarks in electromagnetism.