This repository "Deep Reinforcement Learning Building Energy Control" hosts the source code for a recurrent reinforcement learning agent, specifically tailored for Home Energy Management Systems (HEMS). The agent is trained using a Gym environment based on the CoSES ProHMo Modelica framework. The primary focus of this agent is to efficiently control a building's heat pump and a three-way valve of a thermal storage. The objective is twofold: to adhere to predefined thermal constraints and to optimize the process with a focus on minimizing electricity costs.
Install libraries:
pip install torch numpy pandas gymnasium pyfmi tensorflow
Start policy training:
python ./src/train_rsac.py
- The building simulation was taken from CoSES_thermal_ProHMo Public. Thanks to DZinsmeister for his work.
- Portions of the code, especially those related to RSAC, are adapted from off-policy-continuous-control. Thanks to zhihanyang2022 for his work.
Link to the IEEE Xplore publication
If you use this code or find it helpful for your research, please consider citing our publication:
@INPROCEEDINGS{10202844,
author={Ludolfinger, Ulrich and Zinsmeister, Daniel and Perić, Vedran S. and Hamacher, Thomas and Hauke, Sascha and Martens, Maren},
booktitle={2023 IEEE Belgrade PowerTech},
title={Recurrent Soft Actor Critic Reinforcement Learning for Demand Response Problems},
year={2023},
pages={1-6},
doi={10.1109/PowerTech55446.2023.10202844}
}