Global Optimization of Electromagnets by Random Tabu Search Technique

Abstract

Most of the existing gradient-based iterative techniques for finding the parameters that minimize or maximize an objective function, solely perform local optimization. However, in practical industrial design oriented problems such as selection of the size of an iron-core electromagnet, global optimization will be a key factor in designing a more economical, compact, light and smart product. The application of a global optimization technique would remove the drawbacks of the effect of initial values on local convergence. As the most probable global optimum values irrespective of the starting point are calculated, the random Tabu search method is one of the most powerful solution tools available in recent optimization theory. In our technique, a general finite element based electro-magnetic software MAXWELL is coupled with the Tabu search based optimization routine to numerically determine the final optimum shape of the iron core. The algorithm is illustrated with the help of flow charts and results are presented to verify the efficiency of the algorithm.