Conference Paper

Proceedings of uSim Conference 2024: 4th uSim Conference of IBPSA-Scotland

     

Towards smart despatch of building heating and cooling using artificial intelligence

Qingyuan Wang, Ruosi Zhang, Floris Bierkens, Michael Short, Oleksiy V. Klymenko

Abstract: Improving energy efficiency is one of the most costefficient approaches for decreasing the carbon footprint associated with the built environment as well as reducing the energy bills. This contribution advances the development of smart Building Management Systems able to dynamically respond to the ambient conditions and building usage patterns to minimise energy use for heating/cooling while ensuring the comfort and wellbeing of the users of indoor spaces. In comparison with traditional thermostats or responsive HVAC operation relying on occupancy detection in real time (e.g., using PIR motion sensors) our approach uses a look-ahead strategy employing a hybrid mechanisticmachine learning model. As it collects sensor data on the thermal performance and occupancy patterns of the various spaces within a building, the system improves its ability to predict energy flows between the indoor spaces as well as between the building and its external environment. In combination with short-term weather forecast, this enables our algorithm to minimise the amount of energy used for heating/cooling while ensuring that desired thermal comfort conditions are maintained whenever the space is occupied. The smart look-ahead model thus enables the space to be heated/cooled in anticipation of it being in use while also in many cases switching the heating off before the space is vacated which leads to a reduced overall energy consumption.
Keywords: Energy management, model predictive control, machine learning, artificial intelligence.
Paper: