IoT network-based ANN for ventilation pattern prediction and actuation to optimize IAQ in educational spaces

Abstract

Nowadays, in a user centered design approach, one of the main parameters for assessing the well-being of building spaces is Indoor Air Quality (IAQ), which can assure a crucial level of comfort and optimal conditions to preserve users' productivity and cognitive performance. Research works in this direction mention that with 1000 ppm of CO2 concentration, a reduction of the users' cognitive performance about 11-23% is reported and, for a concentration of 2500 ppm, the decrease reaches 44-94% compared to the performance at 600 ppm. Consequently, a correct buildings ventilation is crucial. The use of mechanical systems seems possibly to avoid the problem but indeed the existing buildings often have outdated and not flexible systems to face changing needs. Thereby, the ventilation rates are not related to people density and the static setup of HVAC systems might be an issue to maintain an acceptable level of CO2 concentration. Moreover, in school buildings, mechanical ventilation is not diffusely adopted and insufficient rates of fresh air supplied to the classrooms are connected with inappropriate IAQ, occurrence of SBS symptoms among pupils. Current technology provides easy measurement of CO2 through dedicated sensors networks. The present research uses the pilot educational building eLUX, located in the Smart Campus of the University of Brescia, to investigate the possibility to integrate IAQ data generated by IoT sensors to improve the estimation of occupancy rate in the educational spaces. The aim is to underline the relevance of the parameter to regulate properly the HVAC systems and to define opening/closing patterns for automated windows to enhance IAQ. The data collected during the monitoring phase are useful to train an Artificial Neural Network (ANN) that through an IoT communication protocol could actuate the ventilation rate control.