2022
- Contingency-constrained economic dispatch with safe reinforcement learning. 2022 21st IEEE International Conference on Machine Learning and Applications (ICMLA), IEEE, 2022 mehr… BibTeX Volltext ( DOI )
M.Sc. Michael Eichelbeck
Technical University of Munich
Informatics 6 - Associate Professorship of Cyber Physical Systems (Prof. Althoff)
Postal address
Postal:
Boltzmannstr. 3
85748 Garching b. München
Place of employment
Informatics 6 - Associate Professorship of Cyber Physical Systems (Prof. Althoff)
Work:
Boltzmannstr. 3(5607)/III
85748 Garching b. München
- Office hours: appointment by email
- Room: 5607.03.035
- michael.eichelbeck@tum.de
Curriculum Vitae
Michael Eichelbeck joined the Cyber-Physical Systems Group as a PhD candidate under the supervision of Prof. Dr.-Ing. Matthias Althoff in October 2021. Previously, he studied control systems at Imperial College London and received his Master’s degree with a thesis on non-cooperative decentralized optimization.
His current research revolves around safe control for power systems by merging reinforcement learning with formal validation. He is a member of the DFG-funded project “Safe-Guarding Artificial Intelligence in Power Systems (SAFARI)“.
Offered Thesis Topics
I am always looking for self-motivated students who are interested in writing a thesis related to my area of research. If you are considering one of the currently offered topics or want to discuss your own research idea, please get in touch via email including your CV, transcript of records, and a brief statement of your motivation.
Currently available
Ongoing/Finished
- [MT] Solving optimal power flow with reinforcement learning
- [BT] Solving optimal power flow with graph neural networks
- [MT] Solving optimal power flow with heterogeneous graph neural networks
- [IDP] Economic dispatch with DC power flow constraints using safe reinforcement learning (co-supervised with Hannah Markgraf)
- [IDP] Machine learning with safety guarantees - Formal conformance checking for prediction models
- [MT] Data Driven Runtime Estimation and Defect Recognition of Batteries in Intelligent Shopping Carts
- [MT] Reinforcement Learning for Intra-Day Energy Trading with Renewable Energy Assets
- [MT] Extension of a German building model approach to multi-national level with the support of machine learning methods
- [MT] Costumer feedback loop for AI-triggered home emergency calls
Teaching
-
Practical Course – Machine Learning for Power Systems (co-organized with Hannah Markgraf)
- SoSe 24 - Reinforcement learning for heat pump control
- SoSe 24 - Smart grid control benchmark scenario designer
- WiSe 23/24 - Forecasting of residential load profiles
- SoSe 23 - Forecasting of wind power generation
-
Practical Course – Verification, Controller Synthesis, and Design of Cyber-Physical Systems
- WiSe 22/23 - Verification of graph neural networks (co-supervised with Tobias Ladner)
-
Seminar – Cyber-Physical Systems
- WiSe 22/23 - Forecasting of renewable energy generation and power demand (co-supervised with Hannah Markgraf)
- WiSe 22/23 - Solving optimal power flow with machine learning (co-supervised with Hannah Markgraf)