Utilizing an Efficient Magnetic Equivalent Circuit Model and Manifold Mapping Method for Two-Level Optimization of Axial Flux Machine

In this paper, a two-level optimization algorithm is developed based on rapid magnetic equivalent circuit (MEC) model and manifold mapping (MM) method. The performance of an axial flux permanent magnet machine (AFPM) is optimized using this approach. The rapid MEC model based on grid division exhibits significant computational efficiency, while reducing the influence of artificial factors in modeling process. Further, quantitative comparison of the error performance under different combinations of divisions are carried out. Subsequently, coarse and fine models are selected, respectively. The iterative process indicates that the modeling approach enables the coarse model to correctly follow the trends of fine model as input changes. As a result, the fast convergence of the MM is successfully achieved.