Data-driven smart control for window and HVAC systems in sports space considering thermal comfort and energy efficiency

Abstract Optimizing the control of window and HVAC systems for thermal comfort and energy efficiency is a critical challenge in intelligent building operations and controls. Despite the application of artificial intelligence of things (AIoT) in supporting smart window and HVAC operations, inaccurate forecasting methods in existing systems hinder their widespread adoption. This study develops a data-driven smart control method, utilizing EnergyPlus-MATLAB co-simulation and Long Short-Term Memory (LSTM)-based time-series forecasting, to clarify the impacts of real-time control of window and HVAC systems on thermal comfort and energy efficiency in large sports buildings. A two-way EnergyPlus-MATLAB co-simulation framework is developed leveraging the Building Controls Virtual Test Bed (BCVTB) platform as the middleware. In addition, a pretrained LSTM model is deployed to quickly predict occupant thermal comfort across large sports spaces, which serves as the basis for window and HVAC control decisions. To demonstrate the feasibility of the proposed approach, a case study of a national fitness hall (NFH) in Shenzhen, China is conducted. The results reveal that the proposed method can maintain more stable thermal comfort though it leads to a 12.26% increase in energy consumption due a 27.9% rise in HVAC operation time. Future work will integrate the proposed method with deep reinforcement learning to further enhance thermal comfort and energy efficiency.