Development of facial-skin temperature driven thermal comfort and sensation modeling for a futuristic application

• Published in: Building and environment Vol. 207; p. 108479
• Main Authors: Jia, Mengqi, Choi, Joon-Ho, Liu, Hanxun, Susman, Gideon
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.01.2022; Elsevier BV
• Subjects: Algorithms; Biosensing; Data analysis; Environmental chambers; Environmental testing; Heat flux; Human-building integration; Infrared detectors; Physical characteristics; Physical properties; Physiology; Predictions; Remote sensors; Skin; Skin temperature; Test chambers; Thermal comfort; Thermal environments; Thermal sensation; Thermoregulation; User-centered model; Human-building integration; Thermal comfort; User-centered model; Thermal sensation; Biosensing
• ISSN: 0360-1323; 1873-684X
• DOI: 10.1016/j.buildenv.2021.108479

The human body is governed by the physiological thermoregulation principle of balancing the heat flux between the body itself and the ambient thermal environment. A number of data-driven thermal comfort assessment approaches have recently been investigated based on the use of real-time sensing over the body, but a potential issue of intrusiveness in sensing locations has also been frequently reported. This research therefore explored the potential use of human facial skin temperature as primary physiological data to develop data-driven thermal comfort and sensation models that would be applied as a futuristic method without worrying about intrusive sensing location issues. This facial skin temperature-driven method has the potential to remove the issue of intrusive locations while providing a user-friendly approach by simply adopting a remote infrared sensor. This study adopted a series of environmental chamber tests with multiple participants recruited to consider the gender ratio in order to identify any significant differences that may be caused by the physical characteristics of human subjects. The average prediction accuracy for thermal comfort in the gradient boosting algorithm is 95.6% and 95.2% for thermal sensation in the data analysis of the first-round tests. Furthermore, data analysis revealed that the model prediction performance was around 80.4% in the validation experiments. Based on the test results, this study developed a facial-skin temperature-driven thermal comfort and sensation model, taking into account individual physiological characteristics, and also identified the most common facial skin areas that provide the best datasets for higher accuracy of comfort/sensation prediction. •The overall thermal sensation (TS) is significantly related to facial skin temperatures.•Facial skin temperatures vary, depending on skin areas and a subject's gender.•TS can be estimated, as a function of facial skin temperatures.•TS prediction acrruracy can vary depending on selected machine learning algorithms.•Among the algorithms, ANN has been found for the most accurate prediction model.