Effects of thermal comfort-driven control based on real-time clothing insulation estimated using an image-processing model

Abstract

The aim of this study is to estimate real-time clothing insulation (R-CLO) and to evaluate the effectiveness of predicted mean vote (PMV)-based control on thermal comfort and electrical energy. For this purpose, an image-processing R-CLO model was developed to estimate the clothing insulation for various ensembles of garments worn by the occupants. The R-CLO model classified 16 individual garments and estimated the total clothing insulation for various ensembles based on these garments. Performance testing using the PMV output from the R-CLO model was conducted. The resulting PMV-based control changed the indoor set temperature according to changes in the clothing insulation, which improved the thermal comfort of the occupants when compared with existing control methods. Even though the proposed control method established a comfortable indoor environment for all clothing conditions, but also affected the electrical energy. The electrical energy is increased as the clothing insulation increased. This study confirmed the potential of comfort-driven control using a vision-based R-CLO model and verified that actual clothing information is required to achieve thermal comfort in the real building as well as to operate the system considering energy.