We keep updating our project🚀. If you like this project, consider giving it a star 🌟 and clicking watch 👀 to keep up with the latest updates and improvements. Your support means a lot!
🔥 BuildingGym is a project that provides an API to easily train reinforcement learning control algorithm for all EnergyPlus environment, and includes implementations of common reinforcement learning algorithm: Policy gradient, DQN, A2C, A3C, and more. Below is the structure for BuildingGym
🎓 BuildingGym is free for students, educators, and academic researchers.
- 😏 Applied to all user-defined EnergyPlus model
- ❤️ Easy implement for common RL algorithms
- 💥 Include common RL algorithms
- 😆 Auto-select the best model
- 😋 Track and visualize all the training process
- 😃 Applied to common control problem, e.g. demand respond, energy saving etc.
Please install this package with python=3.11 and follow Installation guide.docx.
Please note the supported EnergyPlus model file is v23.2. For model in other version, please convert to 23.2. See here for help
The EnergyPlus model in showcase refers to Large Office model in project of Building Energy Models for Commercial Buildings Based on CBECS Data
Please cite our paper:
@article{dai2025buildinggym,
title={BuildingGym: An open-source toolbox for AI-based building energy management using reinforcement learning},
author={Dai, Xilei and Chen, Ruotian and Guan, Songze and Li, Wen-Tai and Yuen, Chau},
Journal={Building Simulation},
pages={1--19},
year={2025},
organization={Springer}
}