Dear all,
We are inviting applicants for the following open position (see below).
With my best regards, Ivan Kondov
PhD Student Position (3 years 75% salary level E13)
Enhancing e-mobility with artificial intelligence and uncertainty quantification
Courses: Mathematics, Computer Science, Physics
Topic The ability to quickly charge car batteries plays a key role in the acceptance of e-mobility. Furthermore, besides fulfilling all of the driver’s requirements, batteries should retain long life times. These challenges in e-mobility are tackled by the battery management system (BMS). To satisfactory meet all requirements, the BMS needs detailed information on battery function. Since this information is limited during operation, the BMS currently works with high safety margins that can significantly reduce battery performance. The plan of this project is to apply machine learning to the battery cells as well as the BMS to achieve an intelligent, individualized cell management that at any point uses each cell at its optimum performance based on extensive lifetime data. Initial training of an AI on a particular battery design is done in the lab using advanced equipment, collecting as much data as possible for subsequent retraining and optimization. In a second step, the BMS AI shall be implemented on simple hardware, where sufficient data is collected and stored such that the AI can be continuously improved individually for each battery cell. One key task of this thesis is to set up a model of the battery by using Gaussian Process Regression. This model (or digital twin) is used by the AI BMS. While providing efficient insight into the dynamics of the battery cell, Gaussian Process Regression enables the use of uncertainty quantification, which is then employed to obtain a more sophisticated idea of the dynamics of the developed model. In order to quantify how the cell behavior is affected by uncertainties, efficient uncertainty propagation models have to be developed and applied to the digital twin.
Requirements Knowledge or experience in at least one of these methodological fields will be an advantage for the candidate: machine learning, artificial intelligence, or uncertainty quantification.
Partners This project is one of 19 AI pilot projects of the Helmholtz Association. It is a joint project of Karlsruhe Institute of Technology, TU München and Forschungszentrum Jülich, and the PhD student will closely collaborate with both partners.
Contacts Prof. Dr. Martin Frank Faculty of Mathematics Steinbuch Centre for Computing Karlsruhe Institute of Technology www.scc.kit.edu martin.frank@kit.edu
Dr. Ivan Kondov Steinbuch Centre for Computing Karlsruhe Institute of Technology www.scc.kit.edu ivan.kondov@kit.edu