Duration: | October 2014 - June 2018 |
Contracting Authority/ Sponsors: | German Federal Ministry of Economic Affairs and Energy (BMWi), Projektträger Jülich, Prozeda GmbH, Steca Elektronik GmbH, Sorel GmbH |
Project Focus: |
Duration: | October 2014 - June 2018 |
Contracting Authority/ Sponsors: | German Federal Ministry of Economic Affairs and Energy (BMWi), Projektträger Jülich, Prozeda GmbH, Steca Elektronik GmbH, Sorel GmbH |
Project Focus: |
The control is an important system component of solar thermal systems. Programming of the control algorithms incurs significant component costs. Here, the use of the neural network methodology could result in significant cost advantages. At the same time, this also offers new possibilities for control optimization with respect to energy consumption optimization. The main objective of the “ANNsolar” project is the development and demonstration of the monetary and technical advantages of the neural network methodology. The option to generate self-learning algorithms enables the realization of complex control strategies for increased energy efficiency in a significantly simpler manner.
In a first step, this project discusses the basics for the application of neural networks in solar thermal systems. The focus is on the development of training and self-learning algorithms. In a second step, a control based on the neural network methodology is realized for the “solar circuit - heating circuit control” application. This should demonstrate the suitability of neural networks for the control of solar thermal heating systems. By using neural networks for the control, energy savings in the order of magnitude of 10-20 % are expected for solar thermal heating systems.
Download project report (only available in german language):
ANNsolar – Künstliche Neuronale Netzwerke für die Anwendung in der Solarthermie