Context
In the context of the energy transition, battery storage systems are playing an increasingly important role for both electric mobility and stationary applications. The EDF Group has positioned itself as a major player both in France and worldwide in this field, thanks to numerous storage projects installed and operated by the EDF Group.

The subject of ageing of Li ion batteries is a major technological issue on which EDF’s R&D has been working for many years. This work is based in particular on a rich history of experimental data collected on dozens of distinct commercial cell references for a total of nearly 15,000 individual tests carried out in EDF’s R&D laboratories. These time series data are centralized on a datalake from which data processing can be carried out with tools adapted to the big data context. Today these data are, among other things, used to build analytical ageing models, but these models only exploit a small portion of the collected data today.

Objectives
The objective of this thesis is to strengthen both our understanding of the ageing of Li ion batteries and our ability to model it by using artificial intelligence. In particular, this thesis sets itself the double objective of using machine learning approaches in order to:

1.      Exploit all the diversity of the experimental database to highlight new links between causes and manifestations of ageing.

2.      Design and test new AI-based ageing models that would bring an advantage compared to classical analytical models. This could for example correspond to the development of light, embeddable and adaptive models, updating their parameters continuously, as new data arrives. One of the final objectives is to optimize the performance and lifespan of batteries according to specific usage conditions.

Organization
This CIFRE thesis will be co-supervised by EDF R&D and the ICube-CNRS laboratory of INSA Strasbourg:

-          Within EDF R&D, experts from the Electrical Equipement Laboratory (Laboratoire des Matériels Electriques - LME) and the Services, Economy, Human Questions, Innovative Tools and AI (Services, Economie, Questions hUmaines, Outils innovants et IA - SEQUOIA) departments will be mobilized to provide both battery and data science expertise.

The ICube laboratory, which is joint laboratory bringing together the CNRS, the University of Strasbourg, ENGEES and INSA Strasbourg.

The PhD student will be working both at the EDF Lab Les Renardières (accessible by public transport from Paris Gare de Lyon and by company shuttle from the surrounding towns) and at INSA Strasbourg.

Students with an engineering degree or a Master 2 with a specialization in Machine Learning or electrical engineering with a minor in data science.

Ideally, the candidate will have a general engineering degree or an initial training in chemistry, physics, electrochemistry, materials science or computer science.

A strong interest in the field of batteries is essential and a first experience on this subject would be a plus.

Note : a second thesis on AI and battery ageing at EDF R&D with a scope complementary to this one will be published soon.