Recent experiences of multiobjective optimisation in electromagnetics: A comparison of methods
-
P. Di Barba [1] ; M.E. Mognaschi [1]
-
[1] 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. 24, Nº 3, 2005, págs. 921-930
- Idioma: inglés
- Enlaces
-
Resumen
Purpose – The aim of this paper is to compare different methods for multiobjective optimisation with respect to the same benchmark problem.
Design/methodology/approach – The following methods are considered: equilibrium of gradients GB, multi‐individual multi‐objective evolution‐strategy MOESTRA, and goal attainment GATT, respectively. They are applied to the shape design of the pole pitch of an electrical machine, with the aim of maximizing the air‐gap induction and minimizing the stray field in the winding.
Findings – The same initial solution of the benchmark is considered for all methods. The final solution of GB and MOESTRA dominates the initial one because both objectives are improved. GB is the most expensive method and therefore is not suited when several non‐dominated solutions should be identified. Accordingly, the trade‐off curve was approximated by means of 15 non‐dominated solutions, resorting to MOESTRA.
Originality/value – A multi‐objective multi‐individual evolution strategy of lowest order has proven to be cost‐effective in identifying a set of non‐dominated solutions, so helping the designer to investigate the structure of the objective space. The requirements in terms of timing and facilities appear to be compatible with the resources of an automated design centre of an industrial company.
