Taguchi-Based Preprocess Strategy for Fast Evaluating Worst Case Cogging Torque Due to PM Tolerance Interactions

Abstract

The Taguchi method can be used with the finite element method (FEM) to evaluate the worst case cogging torque due to permanent magnet (PM) tolerances. However, the sampling number is huge due to interactions of various tolerances. To solve this issue, this article proposes a unique preprocess strategy based on the Taguchi method to significantly reduce the sampling number. By applying the proposed strategy to the 12-slot/10-pole (12S10P) machine, the sampling number is reduced significantly from $4.2 \times 10^{28}$ to 381. Applied to machines with different slot and pole number combinations and compared with existing methods, the proposed strategy is more suitable for machines having small $N_{pm}$ / $C_{T}$ (i.e., $\le 10$ ) and the number of critical PM tolerances (i.e., $\le 4$ ), where $N_{pm}$ is PM number and $C_{T}$ is the greatest common divisor of slot number ( $N_{s}$ ) and pole number ( $2p$ ). The measured results of the 12S10P machine agree well with the finite element predicted results by the proposed preprocess strategy.