Master Thesis Robust Identification of Compositional Electrical Drive Models — Identifiability, Sensitivity & Excitation Analysis

During your thesis, you delve into physics-based models of electric drives and our Python component library while simultaneously reviewing current literature on robust system identification and the design of experiments. Furthermore, you develop innovative robustness diagnostics for our existing identification pipeline—considering parameter identifiability, sensitivity, and convergence behavior—and implement as well as quantify appropriate countermeasures. In Addition, you analyze the excitation content of given datasets, evaluate relevant criteria, and link these insights to the achievable identification quality. You comprehensively evaluate the approach you have developed using a practical benchmark use case. Moreover, you document as well as present your research findings clearly and understandably. Lastly, you extend the approach to address partially observable effects resulting from states that cannot be measured directly, such as temperatures. Education: Master studies in the field of in Cybernetics, Engineering, Mathematics, Computer Science, or comparable Experience and Knowledge: basic knowledge of dynamic systems and differential equations; proficiency in Python programming; understanding of the physics of electrical machines; optional: experience with MATLAB/Simulink for interacting with legacy simulation models; optional: knowledge of machine learning and automatic differentiation frameworks (JAX or PyTorch) Personality and Working Practice: you work independently and systematically on complex issues and possess strong analytical skills, enabling you to accurately grasp and solve problems Work Routine: office attendance required Languages: very good in English