Diffusion-based reinforcement learning for flexibility improvement in energy management systems
- Author
- Siebe Paesschesoone (UGent) , Nezmin Kayedpour (UGent) , Carlo Manna and Guillaume Crevecoeur (UGent)
- Organization
- Abstract
- We introduce an online diffusion-based reinforcement learning approach for enhancing flexibility in energy systems. Leveraging diffusion models, the framework is able to learn the complex dynamics underlying energy systems that are subject to continuous stochastic processes. The flexibility actor, guided by a Q-value based critic, employs a diffusion-based consistency policy, enabling tracing of trajectory points back to initial actions in a single step. The method is applied to optimize the flexibility of grid-connected photovoltaic systems. Results indicate improved learning efficiency and system flexibility compared to standard RL algorithms, including a notable 38.46% increase in final episodic reward, although with a higher computational demand
Downloads
-
(...).pdf
- full text (Published version)
- |
- UGent only
- |
- |
- 1.30 MB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-01JA7P0H27WWFABRJWDXBH7AX7
- MLA
- Paesschesoone, Siebe, et al. “Diffusion-Based Reinforcement Learning for Flexibility Improvement in Energy Management Systems.” 2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE), IEEE, 2024.
- APA
- Paesschesoone, S., Kayedpour, N., Manna, C., & Crevecoeur, G. (2024). Diffusion-based reinforcement learning for flexibility improvement in energy management systems. 2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE). Presented at the 2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE), Dubrovnik, Kroatia.
- Chicago author-date
- Paesschesoone, Siebe, Nezmin Kayedpour, Carlo Manna, and Guillaume Crevecoeur. 2024. “Diffusion-Based Reinforcement Learning for Flexibility Improvement in Energy Management Systems.” In 2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE). IEEE.
- Chicago author-date (all authors)
- Paesschesoone, Siebe, Nezmin Kayedpour, Carlo Manna, and Guillaume Crevecoeur. 2024. “Diffusion-Based Reinforcement Learning for Flexibility Improvement in Energy Management Systems.” In 2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE). IEEE.
- Vancouver
- 1.Paesschesoone S, Kayedpour N, Manna C, Crevecoeur G. Diffusion-based reinforcement learning for flexibility improvement in energy management systems. In: 2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE). IEEE; 2024.
- IEEE
- [1]S. Paesschesoone, N. Kayedpour, C. Manna, and G. Crevecoeur, “Diffusion-based reinforcement learning for flexibility improvement in energy management systems,” in 2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE), Dubrovnik, Kroatia, 2024.
@inproceedings{01JA7P0H27WWFABRJWDXBH7AX7,
abstract = {{We introduce an online diffusion-based reinforcement learning approach for enhancing flexibility in energy systems. Leveraging diffusion models, the framework is able to learn
the complex dynamics underlying energy systems that are subject
to continuous stochastic processes. The flexibility actor, guided
by a Q-value based critic, employs a diffusion-based consistency
policy, enabling tracing of trajectory points back to initial actions
in a single step. The method is applied to optimize the flexibility of
grid-connected photovoltaic systems. Results indicate improved
learning efficiency and system flexibility compared to standard
RL algorithms, including a notable 38.46% increase in final
episodic reward, although with a higher computational demand}},
author = {{Paesschesoone, Siebe and Kayedpour, Nezmin and Manna, Carlo and Crevecoeur, Guillaume}},
booktitle = {{2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE)}},
isbn = {{9798350390421}},
language = {{eng}},
location = {{Dubrovnik, Kroatia}},
pages = {{5}},
publisher = {{IEEE}},
title = {{Diffusion-based reinforcement learning for flexibility improvement in energy management systems}},
url = {{https://attend.ieee.org/isgt-2024/}},
year = {{2024}},
}