User-Adaptable Comfort Control for HVAC Systems

This paper describes a new approach to the control of heating, ventilating, and air-conditioning (HVAC) systems. The fundamental concept of the new approach is that the controller learns to predict the actual thermal sensation of the specific occupant by tuning parameters of a model of the occupant’s thermal sensation. The parameters are adjusted with respect to thermal sensation ratings acquired from the specific occupant and measurements of physical variables that affect thermal sensation so that with time the model accurately reflects the thermal sensation of the specific occupant. From a lumped-parameter model of a singleroom enclosure, it is shown that the stability of the nominal system can be maintained by utilizing a priori information about the parameters of the thermal sensation model. The method is implemented on a ductless, split-system heat pump. Experiments using human subjects verify the feasibility of the method.