Efficient Sampling Algorithm for Electric Machine Design Calculations incorporating Empirical Knowledge
Michael Heroth, Helmut C. Schmid, Wilfried Hofmann · 2022 International Conference on Electrical Machines (ICEM) · 2022
In order to meet the increasing demand for electric vehicles, automotive suppliers such as ZF Friedrichshafen AG are trying to develop modular electric motor platforms. In order to find the optimal platform and to be able to use machine learning techniques, a large amount of data is required. This article shows an overall concept that includes a central database. With this it is possible to build a continuously growing data lake to support the electric machine design with a data-driven solution. The efficient sampling algorithm developed makes it possible to incorporate empirical knowledge into the sampling. In addition, it is possible to reuse calculated designs and no design is calculated more than once. This achieves the efficient generation of large amounts of data, which enable the use of machine learning.