|
چکیده
|
Buildings consume 30%–40% of global energy while simultaneously exposing occupants to poor indoor air quality (IAQ) that impairs health and cognition. This study addresses the challenge of achieving simultaneous energy efficiency, thermal comfort, and healthy indoor environments through intelligent building control. We investigate a hybrid envelope system integrating adaptive façades (operable windows and external shading), a deployable passive daytime radiative cooling (PDRC) roof, and real-time IAQ monitoring (CO2, PM2.5, and VOCs), coordinated by a deep reinforcement learning (DQN) agent balancing three competing objectives. Using validated EnergyPlus–Python cosimulation across four Iranian climates (cold: Tabriz; hot-dry: Yazd; humid: Rasht; and temperate: Shiraz), we compared four configurations: baseline, hybrid envelope with rule-based control, static envelope with DQN, and intelligent hybrid (full integration). In cold and hot-dry climates, the intelligent hybrid system reduced annual energy use intensity by 31% (56.8→38.9 kWh/m2·year in Tabriz and 66.2→45.6 in Yazd), thermal discomfort (PPD) by 70%–75% (35.4%→8.7% in Tabriz and 41.8%→12.5% in Yazd), and mean CO2 levels by 18%–23% (1120–1240→920–950 ppm). Pollution-aware window control reduced PM2.5 exposure during dust storms by 48% versus baseline and 35% versus rule-based control. Decomposition analysis reveals 70% of energy savings derive from envelope technologies, with intelligent DQN control contributing an additional 30% through optimized coordination. Health impact assessment estimates prevention of 5.8 respiratory illnesses and 30 sick days per 250 occupants annually, with 12%–18% cognitive performance improvement valued at 54–81 million IRR per worker/year. Combined benefits yield a simple payback of 2.7–3.2 years. Monte Carlo analysis (500 runs) confirms robustness (CV < 5%). In humid/temperate climates, benefits are reduced but meaningful (18%–22% energy savings). This work demonstrates the first joint adaptive façade-PDRC control under unified deep reinforcement learning with explicit IAQ integration, providing a validated framework for climate-adaptive, health-protective building automation.
|