Estimation of Human Occupancy in Enclosed Spaces Using CO₂ Sensor Data and Statistical Modeling for HVAC Control Applications

This study investigates the estimation of human occupancy in enclosed indoor spaces using CO₂ concentration data obtained from a low-cost sensor. An experimental campaign was conducted in a naturally ventilated university classroom, where CO₂ levels and the number of occupants were recorded under controlled conditions. Statistical analysis, including linear regression and correlation methods, confirmed a strong positive relationship between occupancy and CO₂ accumulation. The selected NDIR sensor (MH-Z16), integrated with an ESP32-based microcontroller, demonstrated sufficient sensitivity and stability for real-time data acquisition. The results validate the use of CO₂ concentration as a reliable indirect indicator of human presence, particularly under standardized conditions with minimal external disturbances. The proposed approach offers a cost-effective and scalable solution for demand-controlled ventilation (DCV) in smart building applications, contributing to energy-efficient HVAC control and proactive indoor air quality and workplace safety management.