Model-based control for BEMS

The stochastic multivariable identification technique is used to model the thermal and moisture behaviour of a full-scale test room with HVAC plant. Models are derived for the unfurnished and furnished conditions and these are shown to predict room temperature and relative humidity, over short and long terms, to a remarkable degree of accuracy. This vindicates the adoption of the time series based identification technique for the modelling of buildings. A robustness assessment shows that the models can adequately cover the range of operating conditions, with the exception of moisture prediction. It is therefore concluded that some degree of moisture-absorbing material is present during the identification phase in order that all the dynamics are properly treated and included in such models.