Master Thesis Data-Efficient Hybrid Machine Learning for Robust Vibration System Prediction

Do you want to bring artificial intelligence into technical applications? In collaboration with a team of engineers and scientists, you will investigate how to develop more robust and reliable predictive models for technical systems. You will work on enhancing a machine-learning toolbox to forecast vibration-loaded systems and add crucial capabilities to learn from real-world insights, especially when measurement data is scarce.

• During your thesis you will research and apply advanced machine learning techniques to integrate limited measurement data into the training of models that currently rely predominantly on simulation data.
• You will develop a benchmark by integrating simulated data and new measurement data from a test bench, utilizing machine learning algorithms to predict the dynamic behavior of nonlinear coupled vibration systems.
• Furthermore, you will apply and evaluate your chosen approaches, comparing their model performance (accuracy and robustness) against simulation-only trained models.
• Finally, you will openly communicate your ideas and contributions, benefiting from the exchange with colleagues within your team, experts in the field, and a broader network across various domains and locations within the company.