Kriging-Assisted Multi-Objective Particle Swarm Optimization of permanent magnet synchronous machine for hybrid and electric cars

In this paper a novel combination of a multiobjective particle swarm optimization algorithm and a Kriging metamodel is presented, which is used for the design optimization of a permanent magnet synchronous machine (PMSM) for electric and hybrid cars. As the evaluation of mostly conflicting objective functions requires computational expensive numerical field simulations, a fast converging optimization algorithm which depends on only a few exact evaluations is needed. Instead of running exact calculations on every design the objective values are estimated by means of a Kriging model in advance and only the most promising design candidates are being computed in a more accurate way. Furthermore, good solutions are searched for directly on the metamodel and particles are spawned at the corresponding positions in the parameter space. The performance of the proposed algorithm is investigated with the help of an analytical test case and used for design optimization of a 10 poles PMSM with 11 parameters and 3 objectives.