Topology optimization is a method of conception which is able to define the topology, the form and the dimensions of a device with the aim of responding optimally to given design criteria. Initially reserved to the sizing of mechanics parts, this method is directed today towards the conception of more complexes devices as those encountered in applied electromagnetic. It is in this context that our work was performed. A topology optimization tool is made of the combination of an optimization algorithm and a material distribution formalism. In a first step, we compared different couplings of metaheuristic algorithms and material distribution formalisms. This comparison led us to choose as optimization tool for the problem under study, the combination of a genetic algorithm and a distribution of material by Voronoi cells. In a second step, we discussed how to improve the exploration and exploitation capabilities of this tool. We have, for this purpose, studied aspects related to the management of the size of the population and to the adaptation of the mechanisms of reproduction to the graphical nature of the problem. After this second step, we builded our optimization tool that we tested on study cases whose complexity is similar to that encountered at industrial showing its potential of to design electromechanical devices.
