On dual control for buildings using persistent excitation condition

Buildings belong to one of the biggest consumers of the final energy. The global effort for saving the energy leads to an intensive research aimed at its optimization. Predictive control has become a very popular approach in many industries with building being no exception. The main bottleneck of this method is, however, a need for a good model. When constructing a building model, it is, in many cases, already operating under the feedback control contradicting thus some of the key assumptions (such as e.g. persistence of excitation or input-noise decorrelation) of a vast majority of the identification approaches. This causes problems for both the initial system identification and model adjustments/reidentification. In latter case, a dual control is a possible approach, when the problems of control and system identification are solved simultaneously. This paper presents an approach when a persistent excitation condition in form of maximization of minimal eigenvalue of information matrix is incorporated into a control criterion.