Neural networks for predictive control of SMRS in industrial energy systems
| dc.contributor.author | Ullmann, Jan | |
| dc.contributor.editor | Mašata, David | |
| dc.date.accessioned | 2025-09-22T08:53:48Z | |
| dc.date.available | 2025-09-22T08:53:48Z | |
| dc.date.issued | 2025 | |
| dc.description.abstract-translated | Decarbonizing energy production and industrial processes is a major global challenge in the eSort to mitigate climate change. Nuclear energy, especially in the form of Small Modular Reactors (SMRs), is increasingly considered as a flexible and low-carbon source of both electricity and process heat. Modern power grids with high shares of renewable energy require energy producers to provide not only stable output but also operational flexibility and ancillary services such as frequency regulation. The operation of SMRs in hybrid systems supplying both electricity and heat presents a complex control problem, as the thermal output must be dynamically split to satisfy variable electricity grid demands and industrial heat requirements. Accurate forecasting and adaptive control strategies are needed to optimize this multi-objective operation while maintaining safety and eSiciency. In our work we show a neural-network-based predictive control framework for SMRs that integrates real grid and weather data to optimize the joint production of electricity and heat. Our tools and results demonstrate that advanced neural network models (LSTM, GRU, Transformer), optimized via multi-objective hyperparameter tuning and supported by attention mechanisms for interpretability, significantly improve forecast accuracy and operational flexibility compared to traditional control methods. This work expands previous knowledge by enabling anticipatory control of SMRs based on real-time system dynamics, thus facilitating their participation in electricity markets and grid ancillary services without compromising thermal supply reliability. The developed software tools and methodology provide a flexible foundation for adapting SMR control to various industrial and grid environments. Overall, this approach supports the integration of nuclear energy into decarbonized, hybrid energy systems and highlights the potential for AI-driven control to enhance the role of SMRs in future sustainable energy infrastructures. | en |
| dc.format | 2 s. | cs |
| dc.identifier.isbn | 978-80-261-1307-2 | |
| dc.identifier.isbn | 978-80-261-1308-9 (printed) | |
| dc.identifier.uri | http://hdl.handle.net/11025/62831 | |
| dc.language.iso | en | en |
| dc.publisher | University of West Bohemia in Pilsen | en |
| dc.rights | © University of West Bohemia in Pilsen | en |
| dc.rights.access | openAccess | en |
| dc.subject | poster | cs |
| dc.subject.translated | poster | en |
| dc.title | Neural networks for predictive control of SMRS in industrial energy systems | en |
| dc.type | konferenční příspěvek | cs |
| dc.type | conferenceObject | en |
| dc.type.status | Peer reviewed | en |
| local.files.count | 2 | * |
| local.files.size | 1620691 | * |
| local.has.files | yes | * |
Files
License bundle
1 - 1 out of 1 results
No Thumbnail Available
- Name:
- license.txt
- Size:
- 1.71 KB
- Format:
- Item-specific license agreed upon to submission
- Description: