Optimizing dynamic energy system for smart buildings using machine learning

Offentlig prosjektsammendrag

Buildings consume a significant percent of the overall energy usage in Norway. It also represents high potential for energy and emissions savings if the right technology is applied. Recent advances in technology has led to the development of ‘smart buildings’ with controls and monitoring systems to improve the performance of the buildings energy infrastructure. The technology is however a passive response system. The application of Artificial Intelligence, specifically Machine Learning, is fast becoming a useful tool in improving complex systems that would otherwise not be improved beyond the current human capacity. Machine learning has found wide applications in autonomous vehicles, computer vision, banking and many other sectors. Machine learning can be used to transform the passive response of smart building to an active system where performance is predicted before time. The integration of machine learning algorithms with an energy system that is fast-response (low temperature system), dynamic (using short term thermal storages), sustainable (using natural working fluid heat pumps and fossil free backup and peak-load energy source) and adaptable will reduce energy and emissions to its avoidable thermodynamic limits. This will reduce both operating and investment cost, improve human comfort levels and increase productivity.