Seasonal effects of thermal comfort control considering real-time clothing insulation with vision-based model

Occupant information is being actively introduced into building control to create a comfortable indoor environment and ensure effective building operation. The occupants' clothing information is a key factor influencing the thermal sensation and must be considered in comfort-based control. Considering these aspects, this study was aimed at comprehensively analyzing the influence of PMV-based control with real-time clothing insulation (R-CLO) on the thermal comfort and system power consumption in different seasons and attires. To this end, a vision-based R-CLO model was advanced, by including stages for person-detection and for garment detection and classification. PMV-based control with the R-CLO model was performed on seven ensembles of winter clothing. The winter experiment results were evaluated and compared with the summer results reported in the previous study to analyze the experimental findings by season. The error of the R-CLO model in estimating the clothing insulation was as low as 0.04 clo. Through PMV-based control considering real-time clothing insulation, the occupants' thermal comfort was enhanced in both summer and winter compared to the existing control methods. Additionally, each 0.1 clo reduction in summer saved average power consumption by 16%, whereas each 0.1 clo increase in winter reduced average power consumption by 13.7%. The existing control strategies appear to prioritize energy over comfort, particularly during the winter. Overall, real-time clothing information can be used for building system control to improve thermal comfort. Furthermore, this study indicates that additional research should be focused on enhancing the system's energy efficiency while appropriately considering the thermal comfort.