The present paper is focused on thermal comfort control problem for building occupants. Thermal comfort is a concept difficult to define and, here, the PMV (Predicted Mean Vote) index is used by means of two predictive control strategies, characterized by having terminal constraints, called here PMV-MBPC (PMV Model Based Predictive Controller). The first thermal comfort control strategy is based on generating a temperature set-point signal that optimizes the building (single zone) internal PMV value. The second one includes the PMV model in the controller prediction computations, generating a non-linear PMV model having Wiener structure. Results related to closed loop system stability are proposed. In this context, an environment for control systems tests is described and the first approach is then implemented in real time using an oil-heater and this environment. Experimental results illustrate the thermal comfort control system performance. Additionally, simulation results, conducted with hourly weather data, also illustrate the control algorithms performance.
Thermal Comfort; PMV; Predictive Control; Optimization
