| Prelegent: | Stephan Scholz, M.Sc., Faculty of Electrical Engineering and Computer Science, Ravensburg-Weingarten University for Applied Sciences, Germany, e-mail: stephan.scholz@rwu.de. |
| Title: | Scientific Machine Learning for Control Engineering |
| Abstract: | In control engineering, we often face two problems: (i) How to model the system to be controlled? (ii) How to design a stable feedback loop? Both questions can be considered as grey box modeling, because both ask for mathematical models with parameters or unknown mathematical functions to be determined. In the proposed approach, differential equations, optimization and machine learning methods are combined to address these problems. As a result, a family of deep neural network models, called neural ordinary differential equations, is introduced and characterized, and its use in control systems design is demonstrated based on selected applications in control engineering. |
