One of the most important and cost-intensive components of a solar thermal plant is the control system. This is why it is so important to develop low-cost technologies in this field. The main task of the solar thermal control system is to identify target and current states in order to ensure the reliable functioning of the entire solar plant and to point to deviations. In particular, the coding process of the control algorithms used here is very time-consuming and therefore very cost-intensive.

Artificial Neural Networks (ANN), by contrast, are to mimic the learning processes in the human brain, so that the complex interactions between the different input variables can be processed quickly. Since insignficant parameters are disregarded, it takes less time to process the data.

In order to test ANN in solar thermal plants, the scientists divided their activities into two steps. First, they developed self-learning algorithms, which – in a second step – were then applied to a “solar circuit - heating circuit control system”. These algorithms could also help to increase the energy efficiency of solar thermal plants by reducing heat loss up to 20%. Both private households and the industry would benefit in equal measure from this.

The research project will continue until June 2018.