Towards practical model predictive control of residential space heating: Eliminating the need for weather measurements

Abstract Model-based control schemes such as model predictive control (MPC) can assist smart-energy systems in achieving higher efficiency and utilization of renewable energy sources. A practical barrier for deploying such control schemes for space heating of residential buildings is the costs related to obtaining the weather data measurements needed for identifying a model that describes the dynamic behaviour of the building. Therefore, this paper reports on a simulation-based study investigating whether there is a significant impact on the performance of MPC schemes when substituting these weather measurements with data from meteorological weather services. Since access to weather forecasts is necessary during the operation of the MPC scheme, this implementation approach draws on data already available to remove the need for weather measurements. The results indicated that this approach only led to a minor performance impact in that heating savings were reduced by 4% while comfort violations increased by less than 0.1 Kh per day on average. The results thereby suggest that the use of data from meteorological forecast services for model identification may constitute a cost-efficient alternative to on-site or near-by weather measurements.