Data-Driven Ventilation and Energy Optimization in Smart Office Buildings: Insights from a High-Resolution Occupancy and Indoor Climate Dataset

This paper explores innovative approaches to reducing energy consumption in building ventilation systems through the implementation of adaptive control strategies. Using a publicly available high-resolution dataset spanning a full year, the study integrates real-time data on occupancy, CO2 levels, temperature, window state, and external environmental conditions. Notably, occupancy data derived from computer vision-based detection using the YOLOv5 algorithm provides an unprecedented level of granularity. The study evaluates five energy-saving strategies: Demand-Controlled Ventilation (DCV), occupancy-based control, time-based off-peak reduction, window-open control, and temperature-based control. Among these, the occupancy-based strategy achieved the highest energy savings, reducing power consumption by 50%, while temperature-based control yielded a significant 37.27% reduction. This paper's originality lies in its holistic analysis of multiple dynamic control strategies, integrating diverse environmental and operational variables rarely combined in prior research. The findings highlight the transformative potential of integrating real-time environmental data and advanced control algorithms to optimize HVAC performance. This study establishes a new benchmark for energy-efficient building management through offering practical recommendations and laying the groundwork for predictive models, renewable energy integration, and occupant-centric systems.