Improved Equivalent Magnetic Network for Optimal Design of Yokeless and Segmented Armature Axial Flux PM Machines

This paper proposes an improved quasi-3D mesh-based equivalent magnetic network (EMN) for optimal design of the yokeless and segmented armature (YASA) axial flux permanent magnet (AFPM) machines. Besides the magnetic field in the air gap, the improved EMN can also predict the armature field in the slots accurately, which is realized by discretizing the armature magnetomotive force (MMF) sources into multiple sub-MMF sources with reference to the distribution of conductors in the slots. The improved EMN can be employed to predict the synchronous inductance, back electromotive force (EMF), electromagnetic torque, and efficiency. Combined with the proposed EMN, a genetic algorithm with elite strategy is applied for the multi-objective optimal design, aiming to improve the YASA-AFPM machine’s torque density and efficiency, with constraints of inverter’s terminal voltage and rated current. Finally, a prototype was manufactured based on the optimized design, and the 3D finite element analysis (FEA) simulation and experiment results verify the accuracy of the proposed EMN and multi-objective optimal design.