Buildings consume a huge amount of energy and mainly utilize it for occupants’ thermal comfort satisfaction. Real-time thermal comfort assessment can enormously contribute to thermal comfort optimization and energy conservation in buildings. Existing thermal comfort models mainly focus on the real-time assessment of occupants’ current thermal comfort. However, in the transient thermal environment, occupants’ thermal comfort is unsteady and varies from time to time. Therefore, if we only assess occupants’ current thermal comfort, prediction error will be elicited. In order to address this problem, it is principally important to comprehend occupants’ real-time thermal sensation trend in the transient thermal environment. This study investigates a novel thermal sensation index that can directly represent an individual’s current thermal sensation trend. By incorporating the novel thermal sensation index into an ordinary thermal comfort model, a composite thermal comfort model is derived, which can simultaneously address an individuals’ current thermal comfort and current thermal sensation trend. Then, by utilizing a machine learning classification algorithm, we propose its intrusive assessment method using skin or clothing temperatures of ten local body parts measured by thermocouple thermometers and its non-intrusive assessment method using a low-cost portable infrared camera. The novel composite thermal comfort model can provide an early warning mechanism for thermal discomfort and contribute to energy conservation in buildings.
Keywords thermal comfort, energy conservation, physiological index, infrared thermography, machine learning