Enabling scalable Model Predictive Control design for building HVAC systems using semantic data modelling

Abstract

Model Predictive Control (MPC) is a promising optimal control technique to reduce the energy consumption of Heating, Ventilation, and Air Conditioning systems in buildings. However, MPC currently involves significant manual efforts in data preparation, control model design, and software interface design. Better semantic representations of buildings, their systems, and telemetry data could help address these challenges. This paper proposes a standard semantic information model and tooling, tailored to BIM software, to streamline MPC design. The approach is tested in an office building, and the generated semantic graph is validated against a use case, where an MPC controller uses Resistance and Capacitance (RC) models that need to be parameterized. The results show that the automatically identified RC models achieve three-hour-ahead temperature predictions for two different rooms within 0.3 °C accuracy. This indicates that semantic data modelling can enable a scalable MPC configuration workflow and more efficient algorithm development and deployment in the future.