Collaboration with Mitsubishi Electric: Improving building control with machine learning

Hybrid control of a set of building systems using data-driven machine learning © Architecture and Building Systems, ETH Zürich

Hybrid control of a set of building systems using data-driven machine learning © Architecture and Building Systems, ETH Zürich

We are pleased to announce the collaboration with our research partner Mitsubishi Electric R&D Centre Europe B.V. for the development of hybrid control algorithms for building systems. 

Control strategies can play a key role in the management of building assets. System operations can be optimized according to a cost function, which can be represented by the indoor environmental quality (IEQ), energy savings, or a combination of metrics. The hybrid approach takes a set of systems (e.g. heating, ventilation, and air conditioning) and searches for operational synergies in terms of energy efficiency and occupant comfort. Most recent control strategies are based either on a model predictive approach or artificial intelligence.

This new research project will investigate the use of data-driven machine learning to enhance the hybrid control of building systems. The algorithm testing will be implemented at the HiLo building, which is a living laboratory based at NEST in Zürich. We look forward to engaging with our partners at the Mitsubishi Electric R&D labs in the UK and Japan over the next four years.  

Research Partner: Mitsubishi Electric R&D Centre Europe B.V. 


Previous
Previous

ETH Energy Blog: Why old buildings matter – The story of building refurbishments

Next
Next

Video: Building Integrated Photovoltaics (BIPV) Workshop 2021