Abstract
Data-driven modeling recently gained interest for various applications within energy systems such as buildings, district heating, or power systems. In some applications, it is used in addition to traditional physics-based simulations; in other cases, physical modeling is replaced by data-driven modeling. A major advantage of data-driven modeling is that data-driven models are less dependent on system parameters and expertise than physical models. However, they are sensitive to data quality. A promising approach is the combination of physical and data-driven models, which allows to use the strengths of each modeling approach. Combined models may then contain multiple sub-models, which can be exchanged during simulation. In our work, a workflow for seamless exchange of sub-models is developed. Our workflow is implemented in Python and Modelica, using a variable-structure model in the Modelica tool Dymola. The variable-structure model is embedded with physical and data-driven sub-models, the sub-models are then exchanged during simulation. Our workflow is applied in a case study of a single-family house heating system.
Original language | English |
---|---|
Title of host publication | Proceedings from the 2nd International Sustainable Energies Conference 2022 |
Place of Publication | Graz |
Publication status | Published - Apr 2022 |
Keywords
- Cyber-Physical Systems
- Simulation and Modeling
- Machine Learning
- Co-Simulation
- Variable-Structure Models