Project 1: Federated Learning Framework for AM
The aim of this project is to use AI methods to correlate input (machine data, materials, part geometries etc.) with output (mechanical properties, geometrical accuracy etc.) along the process chain of an AM process. Expert knowledge shall be integrated in a hybrid AI approach that combines multi-agent systems and federated learning to enable researchers from different institutions to train the AI. This position shall define specifically the framework for the federated learning approach.
Project 2: AM process simulation for AI training data
Artificial intelligence is able to predict AM part performance from input parameters such as material and machine set-up, once sufficiently trained. However, AI relies on big data sets to produce valid results, which can often not be realised in experiments due to high costs and time efforts. The goal of this project is to set up simulation for selected AM process steps in order to simulate training data for the AI.
Project 3: Materials model for AM processes
Metals undergo a complex process chain in Additive Manufacturing, involving phase changes as well as different temperature cycles in the solid state. This results in a complex process – microstructure – property relationship, that needs to be understood along the whole process chain. The goal of this project is to set up a material model for a selected metal alloy that covers all phases and conditions experienced along the process chain, and that can be used in subsequent simulations to predict part properties.