Mechanistic indoor air quality model and data-driven calibration for intelligent HVAC AI control system through real-time data in underground buildings

Abstract

Abstract Modeling the dynamics of indoor air quality (IAQ) in subway environments is challenging due to the complex interplay of variables like subway schedules, ventilation, and passenger numbers. This study developed a high-precision mechanistic model for IAQ management and intelligent HVAC control in underground buildings, focusing on Y-station. Global Sensitivity Analysis (GSA) highlighted the significance of the train piston factor during operational hours and the increased sensitivity of penetration and deposition factors when trains are not operational. The model, calibrated in real-time using a Genetic Algorithm (GA), exhibited remarkable accuracy in simulating PM2.5 levels, affirming its effectiveness in forecasting future air quality. The model adeptly captures the complexities of air quality dynamics, providing a comprehensive understanding of temporal IAQ variations. The result demonstrates the model’s efficacy as a tool offering a foundation for strategies to forecast IAQ and control the HVAC system in underground buildings.