Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : October 15, 2018 - October 16, 2018
Because the performance of permanent magnet synchronous motor (PMSM) is strongly dependent on the quality of magnetization in permanent magnet, the operation for magnetization has to be accurately carried out on permanent magnet. However, the magnetization fault regretfully will occurs more or less. To decrease number of inferior permanent magnet with fault magnetization, identifying the distribution of magnetization in advance is considerably important from the viewpoint of high-quality-manufacturing of PMSM.
One of the methods to identify the magnetization distribution is based on solving inverse problem. Because the number of measurement points of magnetic flux density is different from the number of cells in which the magnetization vector is defined, singular decomposition is additionally required in solving inverse problem. On the other hand, when the target to minimize least square of difference between measurement flux and estimated flux is set, the unconstraint optimization problem can be formulated. Therefore, nonlinear programming can be applied to minimization problem of objective function. In this paper, the identification method based on quasi-Newton method or Newton method is proposed. The numerical performance of proposed method is demonstrated on practical permanent magnet which is employed on interior permanent magnet motor.
