Improving the Accuracy of Spatial Thermal Comfort Assessment through Gappy Data Reconstruction

The health and well-being of building occupants are largely influenced by a comfortable indoor thermal environment. To achieve this, building systems must be controlled based on the occupants' actual thermal comfort needs. Thermal comfort is affected by both environmental and personal factors, and conditions such as building type, system characteristics, and airflow can cause variations in thermal sensation within the same space. Thus, accurately assessing thermal comfort requires the use of environmental data around the occupants. This study aims to predict environmental variables throughout a space from limited sensor data and evaluate spatial thermal comfort. Data reconstruction algorithms, such as Gappy POD and Gappy Autoencoder (AE), were used to restore temperature and air velocity data. DesignBuilder and EnergyPlus simulations were employed for algorithm training and evaluation, with a standard office building as the test case. Results show that using reconstructed data allows for more precise thermal comfort assessments based on location, compared to relying solely on measured data. It was also observed that comfort zones vary by location, highlighting the critical role of precise environmental data predictions for spatial thermal comfort evaluations.