In search for modeling predictive control of indoor air quality and ventilation energy demand in subway station

Abstract

Indoor air quality (IAQ) in underground subway systems is highly influenced by outdoor air pollutants which enter platforms or tunnels through ventilation systems. To eliminate the nonlinear effect of outdoor air quality (OAQ) on IAQ, a predictive model of the neural network and dynamic external analysis is proposed with experimental measuring technical characterization. The partial least squares (PLS) method is used to predict both the particulate matter (PM) concentration and the energy usage of the ventilation system in the underground station. The developed predictive model of ventilation system operation at the D-subway station assured that the outdoor air pollutants do not influence IAQ, since the developed predictive model can remove the effects of OAQ on IAQ in platform as well as the energy used by the ventilation system. A comparison of the results with other prediction models shows that the proposed predictive model can decrease 20% prediction error of PM concentration in platform and also 64% prediction error of energy demand of the ventilation system. These findings suggest that ventilation control strategies should take into account the effects of outdoor air pollutants on IAQ prediction and energy demand.