| Title |
Thermal Comfort Prediction for the Occupant based on Physiological Signals from Wearable Device |
| Authors |
이윤희(Lee, Yoonhee) ; 전정윤(Chun, Chungyoon) |
| DOI |
https://doi.org/10.5659/JAIK.2021.37.10.177 |
| Keywords |
Thermal Comfort Prediction; Wearable Device; Physiological signal; Office; Skin Temperature; Machine Learning Algorithm |
| Abstract |
Thermal comfort is essential to maintain a stress-free environment in a building. This study investigated the thermal environment to develop a
thermal comfort prediction model based on physiological signals and thermal comfort-related responses obtained from a wearable device. Field
experiments conducted in an office during cooling and heating seasons enabled the collection of real-time thermal comfort responses and
physiological signals, such as skin temperature, heart rate, and electrodermal activity of the occupant using the wearable device. We analyzed
the relationships between the thermal comfort responses, physiological factors, and thermal environment to develop an accurate thermal
comfort prediction model. While the skin temperature and electrodermal activity exhibited a significant relationship with the thermal state, a
low heart rate was observed in a more comfortable state. Moreover, machine learning classifiers predicted the thermal comfort state achieved
an accuracy of 80% in both seasons using only physiological data. Thus, the feature importance of the random forest classifier verified that
physiological factors aid the prediction of thermal states significantly. The proposed prediction model can be potentially applied in heating,
ventilation, and air conditioning (HVAC) control. The high performance confirmed the use of wearable devices in identifying the thermal
status of building occupants. |