Does user-centric thermal environment control require real-time recognition of user's clothing condition? A laboratory pilot study

Abstract

Accurate identification of individual thermal states is an important basis for meeting the diverse demands of occupants and achieving user-centric thermal environment control. Developing Personal Comfort Model (PCM) by measuring skin temperature with infrared thermography enables effective thermal sensation prediction. However, existing research is typically conducted under uniform clothing conditions. This experimental design does not consider the differences in clothing between occupants and autonomous clothing adjustment behaviors in real-world scenarios, which limits the application potential of the prediction model based on experimental data. In this study, we conducted experiments in a climate chamber to collect clothing-uncovered skin temperatures (face, neck, arms, and wrists) and subjective evaluation under different clothing insulation (0.35, 0.51, 0.76, 1.01 clo). Statistical analysis and modeling analysis were performed by combining subjective evaluations, skin temperatures, and environmental parameters. The result shows that clothing changes can be reflected in the skin temperature at clothing-uncovered areas, and skin temperature is consistent with thermal sensation. On this basis, the machine learning algorithms were used to evaluate the performance of the thermal sensation prediction model under different combinations of input parameters. The model constructed with artificial neural network algorithm achieved a prediction accuracy of 76.8 % using only nose temperature and air temperature as inputs while clothing information is not included. This represents an approximately 5 % improvement over the PMV model, which requires clothing insulation as an input. This study demonstrates the generalizability of using physiological parameters to predict thermal sensation, providing the theoretical foundation for simplifying individual thermal demand recognition systems.