.png){kind=icon}
.png){kind=icon}
Return
Prediction model for personalized thermal comfort of indoor office workers based on non-skin contact wearable device
.png){kind=icon}
PRE
AI
G
X
J
Y
B
DOI：10.1016/j.buildenv.2025.112686.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Air conditioning control strategies in office environments often fail to account for personal thermal comfort variations among office workers, which may lead to reduced thermal comfort and increased energy consumption. To improve the accuracy of thermal comfort predictions in office buildings, this research proposes a non-skin contact wearable device and an integrated thermal comfort prediction model. The model enhances device convenience by predicting skin temperature and humidity, using them as input parameters for the thermal comfort prediction model. Additionally, the model employs an improved beluga whale optimization (IBWO) algorithm to optimize model parameters and integrates random forest (RF) for feature selection. This approach improves the accuracy of thermal comfort predictions while reducing computational resource demands. The research findings are as follows: (1) Temperature and humidity measurements from wearable device positioned on the chest, shoulders, and back showed greater sensitivity and differentiation compared to measurements from the skin surface, with temperature features generally being more important than humidity. (2) When subjects were in neutral or cool thermal comfort states, their posture tended to be closer to the backrest, causing fluctuations in measurements from the back. (3) By using RF for feature selection and IBWO for model parameter optimization, the average accuracy of thermal comfort prediction improved by 17.04 %, the average F1-score increased by 15.04 %, and the average recall improved by 14.64 %. (4) In real office environment tests, the average accuracy of thermal comfort prediction for the subjects exceeded 98 %.
Keywords:
Non-skin contact
Wearable devices
Indoor office workers
Personalized thermal comfort
Prediction model
Journal
IF:
7.6
Papers:
1.3W
Citations:
6.6W
Organization
Cited Papers