RenoZEB has started to investigate the human-centric monitoring system with an innovative approach developed by the Measurement & Sensor Group at Università Politecnica delle Marche (UNIVPM).
A first test was conducted last January in KUBIK thanks to a joints action of UNIVPM and TECNALIA. Following such an approach, the idea is to get physiological quantities, that are correlated to comfort perception, and use them for the indoor environmental control. The final goal is to obtain a human-based monitoring system that considers the occupants’ side of the comfort problem.
To achieve this objective, a measurement campaign was conducted in January 2020 in KUBIK involving ten participants who were asked to sit inside the monitored test-room while performing light office activities. In the meantime, the room air temperature was varied with step changes from 15°C up to 26°C.
Physiological quantities were collected by means of a smartwatch and also with a thermal camera to have additional information about the heat balance of the user. Data were processed to extract some useful physiological indicators related to human comfort. Then, using a Machine Learning approach, a prediction model was trained to predict the thermal sensation vote of each user with an average root mean square error (RMSE) of 15%.
These indicators could act as driver of an innovative human-centric model: the physiological indicator can be used to measure whether the occupant is experiencing discomfort and, together with ambient parameters, could be used to trigger an actuation on the air temperature set-point to restore comfort conditions.
Written by Nicole Morresi (UNIVPM)