IPMSMの電気的モータパラメータから形状パラメータのベイズ最適化を用いた推定法

抄録

Accurate geometry parameters of interior permanent magnet synchronous motors (IPMSMs) are required to analyze the characteristics of the motor via electromagnetic field analysis. However, in some cases, it may not be possible to fully disassemble the IPMSMs being evaluated, resulting in limited access to certain geometry parameters. This study investigates a method for estimating the geometry parameters of the IPMSMs using Bayesian optimization. The proposed method can estimate ten remaining geometry parameters that cannot be measured from the fourteen parameters and electrical parameters obtained without fully decomposing the motor. Since the composition of the magnets and other materials is unknown, the shape parameters obtained by the proposed method are slightly different from those obtained by actual shape measurements. However, the electrical characteristics are close to those of the actual machine. Even though torque was not used in the estimation, the error could be reduced from 18% to 3.6%.