When designing electric motors, a task which is to be done by many companies (such as Atlas Copco) and which is the motivation of this work, a general nonlinear optimization problem with constraints is to be solved. This problem is characterized by mixed integer parameters and non-standard constraints (which means that the evaluation of some constraints depends on the admissibility of others), and by the use of multi-objective optimization. in this work, we implement a branch-and-bound algorithm which deals with the mixed integer parameters. The nonlinear optimization sub-problems (relaxations) are solved by evolutionary programming. Indeed, the use of classical methods, and in particular the solver Lancelot, turned out to beinappropriate for the solution of this particular problem. After having introduced the problem of modelling electrical motors and the algorithms used for its solution, the results achieved in this work will be presented.
Optimisation en variables mixtes dans le cadre du développement de moteurs électriques
MORIAME, D. (Author). 2009
Student thesis: Master types › Master in Mathematics