Author: Zhang Fan (Hong Kong PolyU) - In indoor environments, demand-controlled ventilation (DCV) is a key technology for achieving efficient air quality control and energy savings. However, the highly uneven distribution of pollutant concentrations poses significant challenges to airflow control strategies. This study proposes a novel airflow control optimization model based on an artificial neural network (ANN). A CFD model is established to simulate and obtain spatial distribution data of pollutant concentrations. Subsequently, kernel density estimation is applied to the frequency distribution of concentrations through statistical analysis, and the ANN is used to learn this uneven distribution, generating a corrected spatial distribution model of pollutants. Finally, this probabilistic model is integrated into an airflow control optimization framework, achieving dynamic optimization of airflow through distribution probability and fan energy consumption models. The results demonstrate that this method effectively quantifies control uncertainty caused by spatial concentration unevenness and significantly reduces fan energy consumption while ensuring indoor air quality. This study provides a new approach for modeling and optimizing uncertainties in DCV system control.
