Motor design can be said as multi-modal optimization problem, as many performances should be considered. In addition, a time-consuming finite element method (FEM) is required for accurate analysis of the motor, and such computational burden becomes worse when the FEM is applied to multimodal optimization problem. In this paper, adaptive-sampling kriging algorithm (ASKA) is proposed to relieve the computation cost of multi-modal optimization problem. The ASKA utilizes kriging interpolation model with generated samples by Compact Search Sampling (CSS) and Exclusive Space-filling Method (ESM). The CSS improves the accuracy of the solutions by generating samples near the expected solutions, and the ESM guarantees the diversity of solutions by generating samples far from existing samples, avoiding solution-near area. Using CSS and ESM, the ASKA adjusts the number of samples effectively and reduces function call considerably. The superior performance of the ASKA was verified by mathematical test functions with complex objective function regions. To validate the feasibility of actual electric machines, the ASKA was applied to optimal design of permanent magnet assisted synchronous reluctance motors for electric vehicles and optimum design with diminished torque ripple is derived. INDEX TERMS Electric vehicles, kriging, multi-modal optimization, optimal design, permanent magnet assisted synchronous reluctance motor (PMa-SynRM)

No abstract

In this paper, subdivided kriging multi-objective optimization (SKMOO) is proposed for the optimal design of interior permanent magnet synchronous motor (IPMSM). The SKMOO with surrogate kriging model can obtain a uniform and accurate pareto front set with a reduced computation cost compared to conventional algorithms which directly adds the solution in the objective function area. In other words, the proposed algorithm uses a kriging surrogate model, so it is possible to know which design variables have the value of the objective function on the blank space. Therefore, the solution can be added directly in the objective function area. In the SKMOO algorithm, a non-dominated sorting method is used to find the pareto front set and the fill blank method is applied to prevent premature convergence. In addition, the subdivided kriging grid is proposed to make a well-distributed and more precise pareto front set. Superior performance of the SKMOO is confirmed by compared conventional multi objective optimization (MOO) algorithms with test functions and are applied to the optimal design of IPMSM for electric vehicle.

No abstract

This paper proposes a modified genetic algorithm that has the same beamforming performance and a fast convergence speed using general genetic algorithm in order to form a beam for the mobile node in a mesh network. The proposed beamforming genetic algorithm selects a part of chromosome a high fitness value in mating process to obtain fast convergence speed, and rest part of chromosome with longer fitness value in order to avoid local solution. Furthermore, the reference beam pattern with Gaussian shape reduces additional convergence speed. Simulation shows that the convergence speed of proposed algorithm improves 20% compared with that of conventional beamforming genetic algorithm.