Occupant information computer vision sensing-based displacement ventilation in large space building for improving indoor environment and energy efficiency

Occupant behavior has great influence on the control and energy efficiency of air-conditioning system. Traditional control methods like preschedule or environment sensor-based PID control could not provide the real-time adjustment for air-conditioning, which may lead to time delay or inappropriate conditioning phenomena, especially in buildings with crowd and large occupancy fluctuation. To address this question, a novel occupant information-based displacement ventilation (OIDV) system is proposed in this research, and the AI powered behavior detection for real-time cooling load calculation and occupant information-based zone displacement ventilation in large space building are first considered. First, a data-driven deep learning (DL) framework (Crowd-YOLO) was trained for occupant count, behavior and position detection and recognition. Then, based on the real-time occupant information sensing results, the displacement ventilation (DV) system was employed to realize cooling air supply flowrate and air supply zones control. Last, the building performance simulation (BPS) for the case study gymnasium which adopted the OIDV system in a 4-day experimental test with 6 different application scenarios was conducted. Results show that the OIDV system could avoid the time delay, as well as the over-conditioning or under-conditioning phenomena of conventional air conditioning control methods. For the optimal case with OIDV system, the indoor temperature could be maintained stable without fluctuation and time delay, the CO2 level was ranged from 450 to 850 ppm, while the outdoor air requirement was reduced by 54.43 %, and total cooling energy consumption was reduced by 50.96 % compared with the baseline case.