in this document, we describe new possibilities offered by genetic algorithms in Electrical Engineering. After analyzing the different existing methods of optimization, we underline their weak and their strong points by comparing them on sorne test problems. The conclusions of this comparative study help us to dçvelop an effective optimization algorithm. This algorithm ensures both a global convergence and low evaluation cost from the function to be optimized. We also consider the fundamental point which consists in introducing sorne supplementary informations concerning the nature of the data of the problem to be treated. This is do ne in order to widen the scope of our optimization problems to various domains of Electrical Engineering. The optimization algorithm was tested and validated on three different applications: - Shape optimization of a cooling structure for Power Electronics component - Optimization of three-dimensional mesh quality for Finite Element software - Shape optimization of an electrornagnetic device based on superconducting coils.
