Research Group CoSES:
The Center for Combined Smart Energy Systems (CoSES) is a low-inertia, sector-coupled, low voltage active distribution grid laboratory to investigates smart operation strategies for sustainable energy systems at distribution grid prosumer levels. The research of our young group covers data driven model validation and calibration, real-time and optimal energy management, and dynamic control strategies of electrical and heat prosumers. The lab hosts the energy infrastructure of five realistic buildings, with real and emulated DERs, and a bidirectional interaction with the electrical and heating networks. The lab uses Power Hardware-in-the-loop (PHiL) philosophy to implement developed control and simulation models in real-time over a real long reconfigurable electrical grid and an emulated heating grid.
Lab website - https://www.mep.tum.de/en/mep/coses/
Research profile - https://shorturl.at/B245c
Topic background:
Data-driven surrogate modelling of physical systems is currently being pushed as real-world application-friendly alternative to extensively detailed equation-based models. Surrogates generally focus on ease-of-use of the model, alignment with realistic data availability, adapting to changing operating conditions and practical abstractions.
Topic description:
In the recent past, we have developed useful surrogates for convex energy system models using LSTM and Transformer architecture. In continuation of this work, we would like to focus on a surrogate for a sector coupled grid-planning problem, which could one-shot predict the total grid import/export demand due to electrification of the mobility and heating sector for any grid topology. Such a surrogate should include the consideration of local energy management systems which would try to maximize self-consumption of PV through demand side flexibility. The surrogate would also require to be validated against a variety of test grids and should be open to transfer learning approaches.
The choice of underlying surrogate modelling principle is not yet finalized. However, from the prior experience a combination of LSTM and transformers, physics-based loss functions and symbolic regression, should be analyzed first.
Tasks:
• Familiarization with the existing surrogate modelling code repository at CoSES
• Analyzing the real-world distribution grid data availability to determine input considerations of the user.
• Building a variety of surrogates using synthetically generated data and validating against real grid datasets.
• Benchmarking the surrogates for performance and robustness
• Results processing, evaluation and presentation
• Preparation of a report describing the key aspects and findings of the conducted work
• Presenting the results to a scientific audience.
• Maintaining a detailed documentation of the code repository
• Participate as a full member of CoSES in all team activities and laboratory duties during your stay.
References:
• Balduin, S. Surrogate models for composed simulation models in energy systems. Energy Inform 1 (Suppl 1), 30 (2018). https://doi.org/10.1186/s42162-018-0053-z
• Niederer, S.A., Sacks, M.S., Girolami, M. et al. Scaling digital twins from the artisanal to the industrial. Nat Comput Sci 1, 313–320 (2021). https://doi.org/10.1038/s43588-021-00072-5
• Cramner, M, Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl, 2023, https://doi.org/10.48550/arXiv.2305.01582
