An algorithm to regulate room temperature

Two researchers from the Urban Energy Systems Lab at the Swiss Federal Laboratories for Materials Science and Technology (Empa) have developed a self-learning algorithm for thermostats. According to an Empa report, the cloud-based solution can be integrated into conventional and smart thermostats and regulate the room temperature with foresight.

“The potential is huge,” said Felix Bünning, co-founder of the Empa spin-off viboo, which is marketing the algorithm. “Our experiments at NEST have shown that this approach can achieve energy savings of between 26 and 49 percent.”

Creating a model of the building requires only two weeks of building data, such as valve positions and room temperature measurements. In combination with predictions of the local outside temperature and global solar radiation, the algorithm then independently calculates the ideal energy input for heating or cooling the building up to 12 hours in advance.

A first partner is the Danish company Danfoss. The international thermostat manufacturer is now implementing the first pilot project with viboo in a conventional building to explore the potential efficiency of the algorithm. The startup is already talking to other industry partners as well. For example, the algorithm will be integrated directly into the building automation system at new offices in Zurich.