Closed-Loop Building Energy Control via Deep Forecasting, Reinforcement Learning and Evolutionary Multi-Objective Optimization in Hot-Summer/Cold-Winter Zon
Abstract
This study proposes a closed-loop building energy control framework for green buildings in hot-summer/cold-winter zones, integrating a three-layer LSTM with attention for short-term load forecasting, a PPO-based reinforcement learning agent for adaptive demand response, and NSGA-II for multi-objective optimization of energy efficiency, comfort, and equipment lifespan. A dataset of 12 office buildings (14 M records over two years) supports training and validation. The forecasting module is evaluated using MAE and RMSE, achieving 6.8% MAE. Comparative experiments with PID, MPC, and single-algorithm baselines show that the proposed method achieves 91.3% energy utilization, an average response delay of 1.9 s, and a comfort compliance rate of 92.4%. Results from both simulation and field deployment confirm the framework’s adaptability and stability under price fluctuations, meteorological disturbances, and multi-building collaboration.References
Boutahri Y , Tilioua A .Reinforcement learning for HVAC control and energy efficiency in residential buildings with BOPTEST simulations and real-case validation[J].Discover Computing, 2025, 28(1):1-26.https://doi.org/10.1007/s10791-025-09544-y.
Wei T , Wang Y , Zhu Q .Deep Reinforcement Learning for Building HVACControl[J].ACM,2017.https://doi.org/10.1145/3061639.3062224.
Gao G, Li J, Wen Y. EnergyEfficient Thermal Comfort Control in Smart Buildings via Deep Reinforcement Learning[J]. arXiv preprint, 2019, arXiv:1901.04693.https://doi.org/10.48550/arXiv.1901.04693.
Eini R, Abdelwahed S. Learningbased Model Predictive Control for Smart Building Thermal Management[J]. arXiv preprint,2019,arXiv:1909.05331.https://doi.org/10.48550/arXiv.1909.05331.
Lim SH. Robust deep reinforcement learning for personalized HVAC system[J]. Energy and Buildings, 2024, 319:114551.https://doi.org/10.1016/j.enbuild.2024.114551.
Sayed K A , Boodi A , Broujeny R S ,et al.Reinforcement learning for HVAC control in intelligent buildings: A technical and conceptualreview[J].Journal ofBuildingEngineering,2024,95.https://doi.org/10.1016/j.jobe.2024.110085.
Manjavacas A , Campoy-Nieves A ,Jiménez-Raboso, Javier,et al.An experimental evaluation of Deep Reinforcement Learning algorithms for HVAC control[J].Artificial Intelligence Review,2024.https://doi.org/10.1007/s10462-024-10819-x.
Yu J , Schreck J S , Gagne D J ,et al.Reinforcement Learning (RL) Meets Urban Climate Modeling: Investigating the Efficacy and Impacts of RL-Based HVACControl[J].2025.https://doi.org/10.48550/arXiv.2505.07045.
B Z W A , A Z Z , B B L A ,et al.Residential heating energy consumption modeling through a bottom-up approach for China's Hot Summer–Cold Winter climatic region[J].Energy and Buildings, 2015, 109(Dec.):65-74.https://doi.org/10.1016/j.enbuild.2015.09.057.
Tong, Gui Q .Adaptability Analysis of Passive Building Energy Efficiency Technology in Hot Summer and Cold Winter Region[J].Applied Mechanics & Materials,2013,409-410:651-654.https://doi.org/10.4028/www.scientific.net/AMM.409-410.651.
Merabet GH, Essaaidi M, Ben Haddou M, et al. Intelligent building control systems for thermal comfort and energy-efficiency: A systematic review[J]. arXiv preprint, 2021,arXiv:2104.02214.https://doi.org/10.48550/arXiv.2104.02214.
Hanumaiah V, Genc S. Distributed multi-agent deep reinforcement learning framework for whole-building HVAC control[J]. arXiv preprint, 2021, arXiv:2110.13450.https://doi.org/10.48550/arXiv.2110.13450.
Hosseinloo AH, Ryzhov A, Bischi A, et al. Data-driven control of micro-climate in buildings: Event-triggered RL approach[J]. arXivpreprint,2020,arXiv:2001.10505.https://doi.org/10.48550/arXiv.2001.10505.
Ghahramani A. Artificial intelligence for efficient thermal comfort systems[J]. Frontiers in Built Environment, 2020, 6: 49.https://doi.org/10.3389/fbuil.2020.00049.
Ogundiran J. A systematic review on the use of AI for energy efficiency in buildings[J]. Sustainability, 2024, 16(9): 3627.https://doi.org/10.3390/su16093627.
Jain A , Smarra F , Reticcioli E ,et al.NeurOpt: Neural network based optimization for building energy management and climate control[J].arXiv, 2020,arXiv:2001.07831.https://doi.org/10.48550/arXiv.2001.07831.
Jiang Z, Risbeck MJ, Ramamurti V, et al. Building HVAC control with reinforcement learning for reduction of energy cost and demand charge[J]. Energy andBuildings,2021,239: 110833.https://doi.org/10.1016/j.enbuild.2021.110833.
Xu S , Fu Y , Wang Y ,et al.Efficient and assured reinforcement learning-based building HVAC control with heterogeneous expert-guided training[J].SCIENTIFIC REPORTS,2025,15(1).https://doi.org/10.1038/s41598-025-91326-z.
Henze G , Schoenmann J .Evaluation of Reinforcement Learning Control for Thermal Energy Storage Systems[J].Hvac &RResearch,2003,9(3):259-275.https://doi.org/10.1080/10789669.2003.10391069.
Kurte K , Munk J , Kotevska O ,et al.Evaluating the Adaptability of Reinforcement Learning Based HVAC ControlforResidentialHouses[J].Sustainability,2020,12(18):7727.https://doi.org/10.3390/su12187727.
DOI:
https://doi.org/10.31449/inf.v49i14.11233Downloads
Published
How to Cite
Issue
Section
License
Authors retain copyright in their work. By submitting to and publishing with Informatica, authors grant the publisher (Slovene Society Informatika) the non-exclusive right to publish, reproduce, and distribute the article and to identify itself as the original publisher.
All articles are published under the Creative Commons Attribution license CC BY 3.0. Under this license, others may share and adapt the work for any purpose, provided appropriate credit is given and changes (if any) are indicated.
Authors may deposit and share the submitted version, accepted manuscript, and published version, provided the original publication in Informatica is properly cited.
