Explore projects

In this project, the current state of research on the application of neural networks regarding topology-optimizaion is reviewed. Based on this, an experimental tool is developed that implements Sosnovik and Oseledets’ approach to AI-supported topology optimisation. A convolutional neural network is used here, which optimises structures based on initial iterations using training data. The developed programme uses Python and the Keras deep-learning-library. The aim is to use AI to reduce the computational effort of topology optimisation without compromising accuracy. The results show that neural networks reveal potential to significantly accelerate optimisation processes, especially for recurring, complex problems. Nontheless there are challenges in generalizing the methods to appropriately handle the aimed range of problems.

Model generator for the optimization of energy systems, based on the modelling framework oemof. No programming skills are required to use it, as the parameters are entered using spreadsheets (e.g. Excel, LibreOffice).

Separate Deployment-Einheiten

Microservices mit Docker