PhD Position in Physics-informed Machine Learning for Railway Infrastructures

Updated: about 1 month ago
Deadline: 02 Jun 2021

The booming advance of artificial intelligence (AI) provides us with unprecedented opportunities in shaping our digital era. In the field of structural engineering, AI plays an important role in digital transformation concerning the whole life cycle of structures. However, one remaining challenge in AI research is the lack of interpretability, as most AI applications are driven only by data while neglecting the underlying physics. Until now, we have discovered and created knowledge for an in-depth understanding of the physics behind the functioning of engineering structures. Creating AI that can understand and utilize this knowledge is crucial for enabling better solutions for practical problems in engineering structures. Potential applications include model parameter identification, structural health monitoring, maintenance decision optimization, structural design/upgrade, and digital twins.

In this PhD project, you will explore and develop physics-informed machine learning methods for modelling the structural mechanics and dynamics of railway infrastructures. The developed methods will be rooted in state-of-the-art mathematical models of railway infrastructures, aiming to represent our existing knowledge by machine learning with global or local interpretability. As part of this project, you will engage in discussions with railway industry partners and receive feedback. Limited teaching responsibilities are also included in this position.

As a PhD student, you will be a member of the Railway Engineering section in the Department of Engineering Structures at the Faculty of Civil Engineering and Geosciences, TU Delft. You will work closely with ProRail, the Dutch railway infrastructure manager, on the validation and implementation of your research project and potential applications. The supervisor team consists of Prof. Rolf Dollevoet (promotor) and Dr. Hongrui Wang (co-promotor).

View or Apply

Similar Positions