Source: Energy & Management Powernews, March 22, 2022
A project at the Fraunhofer Institute for Solar Energy Systems is investigating how the operation of heat pumps can be adapted to changing conditions.
Whether the heat pump can actually fulfill the role often attributed to it as the heating technology of the future depends to a large extent on its efficiency. But this does not always meet the expectations of operators in practice, says a statement from the Fraunhofer Institute for Solar Energy Systems (ISE). Therefore, scientists there are developing a new generation of intelligent heat pumps together with partners from industry and research. Artificial intelligence (AI) is to enable adaptation to changing conditions. An energy saving up to 20% and a corresponding CO2 reduction are attainable in this way without comfort losses, it is further said.
The scientists point out that with conventional heat pumps in residential buildings the heating curve control is set by the installer in each case according to the desired room temperature and the outside temperature. Such static control does not take into account, for example, changing user habits or the aging of the building. "In this context, the progressive digital networking of heat pump and building via smart home sensor technology has created the prerequisite for self-learning adaptive control," explains Lilli Frison, project manager at Fraunhofer ISE.
Focus also on user comfort and predictive maintenance
New AI methods based on artificial neural networks should enable adaptive heat pump control and monitoring. The heat pump manufacturer Stiebel-Eltron, the French energy company EDF, and French research institutes are working on this with Fraunhofer ISE in a three-year project funded by the German Federal Ministry of Education and Research.
In addition to the highest possible energy efficiency, the partners also have user comfort in mind with the new generation of heat pumps. In addition, artificial intelligence is to ensure predictive maintenance and help avoid performance losses through early fault detection. "To this end, novel machine learning methods from the research field of 'incremental learning' are to be developed so that the AI is quickly operational and continues to learn autonomously and adaptively during operation," says project manager Frison. This is how the predicted 20 percent energy savings should actually be achieved.
The ultimate goal of the project is "to be able to transfer the AI methods cost-effectively to a large number of different building types, in which the data also often comes from sensors with low precision and reliability," according to the Fraunhofer Institute's statement. It said the focus is on the use cases "adaptive heating curve control," "adaptive control of hot water heat pumps based on load forecasts" and "adaptive fault detection and diagnosis." In a next step, these heat pump controls will be simulated in a test environment and then validated in laboratory tests and pilot buildings.
Author: Fritz Wilhelm