Adaptive predictive control of thermo-active building systems (TABS) based on a multiple regression algorithm

There is a growing trend toward the use of thermo-active building systems (TABS) for heating and cooling buildings, since these systems are known to be very economical and efficient. However, due to the large thermal inertia, the control of these systems turns out to be complicated and time-consuming in the parameterization. With standard control strategies, the required thermal comfort in buildings often cannot be met, especially if the internal heat sources are suddenly changed. This paper presents a novel adaptive and predictive computation method, based on multiple linear regression (AMLR) for the control of TABS. Through the self-learning effect, no parameterization of heating and cooling curves is necessary. The algorithm is compared to a conventional control strategy for TABS. In this simulation case, it turns out that the adaptive and predictive control strategy can achieve a saving of pump-running time of up to 81% while increasing thermal comfort. In addition, the algorithm is validated by a three-week laboratory experiment. With simple calculation approaches, this algorithm proves to be very practical and can easily be integrated into a building's automation system. Furthermore, it offers the possibility for load shifts from the use of the thermal mass of a building.