User-centric approach to optimizing thermal comfort in university classrooms: Utilizing computer vision and Q-XGBoost reinforcement learning

The ’one-size-fits-all’ setting of heating, ventilation, and air conditioning (HVAC) systems in university classrooms not only compromises occupant comfort but also leads to potential energy wastage. This study proposes a user-centered approach to optimize the thermal environment in university classrooms by integrating computer vision and Q-XGBoost learning. Computer vision facilitates non-intrusive and real-time sensing of thermal comfort in dynamic university classroom environments. Simultaneously, the Q-XGBoost model identifies optimal operations for HVAC systems by learning from the interactions between occupants and their environment, ensuring a balance between comfort and energy efficiency. These innovative methodologies are harmoniously integrated into a smart air conditioning control box (SACC-Box), achieving intelligent operation of HVAC systems in the real world. Subsequently, a controlled study was conducted to test the user-centered approach's performance in improving thermal comfort and energy efficiency. The results reveal that classrooms equipped with the SACC-Box significantly improve occupant satisfaction with thermal comfort by approximately 13.5 % while concurrently achieving up to a 5 % reduction in energy consumption compared to classrooms operating with traditional built-in smart control systems. This investigation offers a viable solution to balance energy efficiency and user comfort in university classrooms, underlining the transformative potential of integrating intelligent technologies that inspire a greener, more adaptable, and user-centric futures in higher education establishments.