Electromagnetic Optimal Design of a PMSG Considering Three Objectives and Using NSGA-III

This article presents the optimal design of a permanent magnet synchronous generator (PMSG). A finite element (FE) model is used to construct a metamodel, which afterward is utilized to define the objective function that models the PMSG. Kriging modeling is employed along with the design of experiments based on Latin hypercube sampling. The utilization of a surrogate model allows to speed up the optimization process while keeping the accuracy since they are developed from the FE analysis. On the other hand, it has been reported that the non-sorting genetic algorithm (NSGA) III is better than NSGA-II because it can solve multi- and many-objective optimization problems. This article demonstrates by numerical experiments that NSGA-III can be successfully used in the optimal design of PMSG with three objectives.