RC building modelling for control purposes: A case study

When dealing with models, a key factor to consider when selecting their features is the context in which the models will be used: for example, they could be used for design or for control purposes. If we focus on the second case, the model should be accurate enough to capture the principal dynamics of interest and simple enough to minimize the computational effort. In building modelling for control, a promising paradigm seems to be the use of simplified grey-box models. This paper presents a case study in which the existing temperature control strategy can be improved with the resulting possibility of considerable energy saving. More in detail, we introduce here the first step of the entire process: the choice of the model of the system. We decided to investigate the use of a grey-box model, the parameters of which were estimated using a parametric identification process. Thanks to this approach, full knowledge of the system is not required but this lack of information needs to be balanced with the use of measured data. We decided to use only measured data during the standard operation mode of the system for the parameter identification process. Thus we did not perform targeted experiments on the real system, because of all the restrictions in the specific context. Using this approach, it was still possible to achieve good results in terms of deviation between model simulation and data (indoor air: RMSE = 0.31 and R2 = 0.92).