Daniel Bayer - Chair of Computer Science III

Phone	(0931) 31-89632
Telefax
E-Mail	daniel.bayer@uni-wuerzburg.de
Room	01.019
Address	Lehrstuhl für Informatik III Am Hubland 97074 Würzburg Germany

Research Interests

Daniel Bayer works as a research assistant at the Chair of Computer Science (Modelling and Simulation). He first studied mathematics with a Bachelor of Science degree, then computer science at the Friedrich-Alexander-Universität Erlangen-Nürnberg.
His work focuses on the research project of DigiSWM. Here, one of the central goals is to combine extensive energy data with AI methods to simulate and evaluate new energy services.

In particular, his research interests focus on the following topics:

data-driven or agent-based simulations, especially in the energy & building sector
data analysis of time series
multi-agent reinforcement learning
sustainable computing

Research projects

Recent Publications

2023[ to top ]

A digital twin of a local energy system based on real smart meter data Bayer, Daniel; Pruckner, Marco in Energy Informatics (2023). 6(1)
- [ BibTeX ]
- [ EndNote ]
@article{2023_Bayer_DT_LocalEnergySystem, author = {Bayer, Daniel and Pruckner, Marco}, journal = {Energy Informatics}, keywords = {pruckner}, month = {mar}, number = 1, publisher = {Springer Science and Business Media {LLC}}, title = {A digital twin of a local energy system based on real smart meter data}, volume = 6, year = 2023 }

%0 Journal Article %1 2023_Bayer_DT_LocalEnergySystem %A Bayer, Daniel %A Pruckner, Marco %D 2023 %I Springer Science and Business Media {LLC} %J Energy Informatics %N 1 %R 10.1186/s42162-023-00263-6 %T A digital twin of a local energy system based on real smart meter data %V 6

2022[ to top ]

Enhancing the Performance of Multi-Agent Reinforcement Learning for Controlling HVAC Systems Bayer, Daniel; Pruckner, Marco in 2022 IEEE Conference on Technologies for Sustainability (SusTech) (2022). 187–194.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ URL ]
Systems for heating, ventilation and air-conditioning (HVAC) of buildings are traditionally controlled by a rule-based approach. In order to reduce the energy consumption and the environmental impact of HVAC systems more advanced control methods such as reinforcement learning are promising. Reinforcement learning (RL) strategies offer a good alternative, as user feedback can be integrated more easily and presence can also be incorporated. Moreover, multi-agent RL approaches scale well and can be generalized. In this paper, we propose a multi-agent RL framework based on existing work that learns reducing on one hand energy consumption by optimizing HVAC control and on the other hand user feedback by occupants about uncomfortable room temperatures. Second, we show how to reduce training time required for proper RL-agent-training by using parameter sharing between the multiple agents and apply different pretraining techniques. Results show that our framework is capable of reducing the energy by around 6% when controlling a complete building or 8% for a single room zone. The occupants complaints are acceptable or even better compared to a rule-based baseline. Additionally, our performance analysis show that the training time can be drastically reduced by using parameter sharing.

@inproceedings{9794179, abstract = {Systems for heating, ventilation and air-conditioning (HVAC) of buildings are traditionally controlled by a rule-based approach. In order to reduce the energy consumption and the environmental impact of HVAC systems more advanced control methods such as reinforcement learning are promising. Reinforcement learning (RL) strategies offer a good alternative, as user feedback can be integrated more easily and presence can also be incorporated. Moreover, multi-agent RL approaches scale well and can be generalized. In this paper, we propose a multi-agent RL framework based on existing work that learns reducing on one hand energy consumption by optimizing HVAC control and on the other hand user feedback by occupants about uncomfortable room temperatures. Second, we show how to reduce training time required for proper RL-agent-training by using parameter sharing between the multiple agents and apply different pretraining techniques. Results show that our framework is capable of reducing the energy by around 6% when controlling a complete building or 8% for a single room zone. The occupants complaints are acceptable or even better compared to a rule-based baseline. Additionally, our performance analysis show that the training time can be drastically reduced by using parameter sharing.}, author = {Bayer, Daniel and Pruckner, Marco}, booktitle = {2022 IEEE Conference on Technologies for Sustainability (SusTech)}, keywords = {pruckner}, month = {April}, pages = {187-194}, title = {Enhancing the Performance of Multi-Agent Reinforcement Learning for Controlling HVAC Systems}, year = 2022 }

%0 Conference Paper %1 9794179 %A Bayer, Daniel %A Pruckner, Marco %B 2022 IEEE Conference on Technologies for Sustainability (SusTech) %D 2022 %P 187-194 %R 10.1109/SusTech53338.2022.9794179 %T Enhancing the Performance of Multi-Agent Reinforcement Learning for Controlling HVAC Systems %U https://ieeexplore.ieee.org/document/9794179/ %X Systems for heating, ventilation and air-conditioning (HVAC) of buildings are traditionally controlled by a rule-based approach. In order to reduce the energy consumption and the environmental impact of HVAC systems more advanced control methods such as reinforcement learning are promising. Reinforcement learning (RL) strategies offer a good alternative, as user feedback can be integrated more easily and presence can also be incorporated. Moreover, multi-agent RL approaches scale well and can be generalized. In this paper, we propose a multi-agent RL framework based on existing work that learns reducing on one hand energy consumption by optimizing HVAC control and on the other hand user feedback by occupants about uncomfortable room temperatures. Second, we show how to reduce training time required for proper RL-agent-training by using parameter sharing between the multiple agents and apply different pretraining techniques. Results show that our framework is capable of reducing the energy by around 6% when controlling a complete building or 8% for a single room zone. The occupants complaints are acceptable or even better compared to a rule-based baseline. Additionally, our performance analysis show that the training time can be drastically reduced by using parameter sharing.

Daniel Bayer M. Sc.

Research Interests

Research projects

Recent Publications

Data privacy protection

Data privacy protection

Picture credits