Development of a novel method to detect clothing level and facial skin temperature for controlling HVAC systems

Abstract

People spend most of their time indoors, and thus it is important to provide occupants with a comfortable indoor thermal environment. However, inappropriate thermostat temperature settings in offices make occupants less comfortable. This study developed a new control strategy for HVAC systems that adjusts the thermostat setpoint according to clothing level and mean facial skin temperature. An image-classification model was trained on the basis of a convolutional neural network (CNN) to classify the clothing level of occupants, which was then used to calculate a comfortable air temperature. This investigation used a long-wave infrared (LWIR) camera with a face-detection program to obtain occupants’ mean facial skin temperature. This study performed experimental tests to correlate mean facial skin temperature with thermal sensation votes. The mean facial skin temperature was then used to develop a control strategy for an HVAC system in a single-occupant office. With the use of the control strategy, 91% of the subjects tested in this investigation felt thermally neutral in the office.