Multiobjective Shape Optimization of Segmented Pole Permanent-Magnet Synchronous Machines With Improved Torque Characteristics

Magnet segmentation is an effective and simple technique for cogging torque reduction in high power permanent-magnet (PM) synchronous machines; however, it deteriorates air gap flux density and decreases the output torque. Therefore, a multiobjective optimization framework is necessary for cogging torque minimization, and to diminish its adverse effect on the output torque in segmented-pole permanent-magnet synchronous machines (PMSMs). This can be fulfilled by proper selection of widths and displacements of the magnet segments. Finite-element analysis (FEA) is an accurate method for this purpose. However, it is very time consuming where finding optimal configuration needs a lot of simulations. Thus, an analytical based design optimization is very useful and eases the design process. In this paper, a novel semianalytical model for cogging torque computation in PMSMs is proposed. Based on the proposed model, a multiobjective optimization framework is developed. The particle swarm optimization (PSO) method is applied to find the optimum machine design. To show the effectiveness of the proposed method, two prototype segmented magnet PMSMs with two and three PM blocks per pole are optimized respectively. Performance characteristics are compared to the initial machine design and segmented PMSMs with design parameters chosen according to previous analytical models and initial uniform pole machines using FEA.