Variable Mesh Adaptive Direct Search Algorithm Applied for Optimal Design of Electric Machines Based on FEA

Optimal design of electric machine based on finite element analysis (FEA) is characterized with the nonlinear magnetic saturation, which causes the much longer computation time for maintaining the high accuracy. In order to compensate the excessive computation time due to FEA and guarantee the reliable convergence to global optimum, variable mesh adaptive direct search (VMADS) is newly implemented with multi-start in this paper. Effectiveness of VMADS has been verified through a benchmark function. In particular, the proposed algorithm has been employed to optimal design of interior permanent magnet synchronous machine (IPMSM) for maximum torque per ampere (MTPA), of which design objective should be evaluated by FEA for better accuracy.