Supervision: Benjamin Ricaud

Project type: Master thesis


This project is in collaboration with the Paul Scherrer Institute. The goal is to better understand the mechanical phenomena appearing in new generation superconductors designed for future CERN experiments in particle physics. The constraints on these materials are enormous due to high currents and magnetic fields and it is important to predict their failure before it actually happens. Sensors (acoustic, voltage, current) record data related to the state of the material. Various events, with different signatures are visible and could allow for a prediction of the failure as well as a characterisation of the state of the material.

From these sensor data the goal is to use supervised as well as unsupervised machine learning to classify the different events occuring and predict failures. The student will test a wide range of ML methods and possibly design new ones adapted to the dataset of sensor measures. Sensor data are rather different from the usual image, sound or text datasets. A mixed approach involving signal processing methods combined to machine learning will be adopted to help the machine detect the most relevant information, be more accurate and learn faster.

The student must have a good knowledge of signal processing. Previous projects and experience working on the analysis/filtering of time series and/or sensors signals and machine learning would be a plus. A basic knowledge of machine learning is required and the student is expected to learn advanced ML techniques during the project.