A Systematic Design Methodology Based on Data Clustering for Automotive Drive Cycle Oriented Optimization of Electrically Excited Synchronous Machines

Traction motors for electric vehicles are typically optimized towards minimum material cost, which is directly connected to the motor weight and volume as well as the loss energy in an automotive drive cycle. Thus, design and optimization of traction motors for electric vehicles require accurate knowledge of the loss energy in an automotive drive cycle. As most drive cycles consist of hundreds or thousands of operating points, reducing the number of operating points by utilizing data clustering based methods can save considerable computation time. On the other hand, the less drive cycle points considered, the lower the accuracy of the calculation of loss energy. In this paper, k-means data clustering and energy center of gravity method are investigated regarding their accuracy for a varying number of clusters and different types of drive cycles. While the focus of this study lies on the electrically excited synchronous machine (EESM), the data is presented for the permanent magnet synchronous machine (PMSM) as the prevalent traction motor type as well. A comparison between both machine types is made to highlight the different requirements of the EESM compared to the PMSM.