Energy-Efficient Zonal HVAC Control using Iot, CCTV Occupancy Analytics and MPC

Buildings account for a substantial share of global energy demand and greenhouse-gas emissions, with HVAC systems often the largest loads in operation. At the same time, indoor air quality (IAQ) influences cognition, comfort, and health - typically managed by limiting CO2 concentrations below regulatory or de-facto thresholds. Traditional timetable-based or rule-based ventilation struggles to track rapidly varying occupancy, causing IAQ excursions and energy waste. Leveraging IoT sensors and CCTV analytics to obtain real-time occupancy signals to correlate with indoor environment variables, together with predictive control, enables proactive ventilation that better balances IAQ and energy. This paper presents a novel occupancy-aware predictive HVAC framework that fuses real-time occupant estimates from CCTV analytics with a disturbance-aware ARX model for CO2 concentration (recursively identified from measured data), along with a real-time, energy-aware MPC scheme. The MPC includes a cubic-law fan power model and a dynamic minimum-ventilation constraint derived from static designed occupancy (per-area and per-person outdoor-air requirements) and zonal airflow characteristics. Using measured hourly occupancy estimates and CO2 data from the Net Zero Industry Innovation Centre (NZIIC) in Middlesbrough, UK, we identify an accurate ARX(2,2) model and use it in Hardware-in-the-Loop (HIL) experiments to validate the approach. Results demonstrate that the controller varies airflow predictively and in step with occupancy, maintaining optimal Indoor Air Quality (IAQ) well under a 1000 ppm limit, while substantially reducing fan energy requirement by 85% versus a static preprogrammed ventilation strategy designed for average designed daytime occupancy with night setback as a comparative baseline.