Real-time ventilation control for indoor CO₂ management using occupant information

In contemporary Conventional ventilation control methods primarily rely on environmental measurements taken indoors but often overlook occupant-specific factors. This study presents an optimal ventilation control algorithm based on a real-time indoor CO₂ concentration prediction model designed to enhance IAQ and energy efficiency. This model incorporates real-time occupant data, highlighting the considerable influence of occupant-related variables on the indoor CO₂ levels. To this end, three deep learning architectures—deep neural networks, long short-term memory (LSTM), and gated recurrent units—were evaluated, with the LSTM model exhibiting superior accuracy and robustness. Using this model, a predictive ventilation control algorithm was developed to proactively regulate airflow and maintain CO₂ concentrations below the recommended threshold of 1,000 ppm. The effectiveness of the proposed control strategy was validated using mockup experiments and living lab-based simulations. The results show that integrating real-time occupant data considerably enhances indoor comfort than rule-based ventilation control. Furthermore, optimal ventilation control resulted in a considerable decrease in energy consumption by approximately 24.74%, particularly in large-scale environments. These findings highlight the potential of the proposed method as a robust solution for next-generation indoor environmental management systems and intelligent control in smart buildings.