• Funding : ANR, Artois
  • Start year :
  • 2023

Abstract

In the midst of the ongoing energy transition, lithium-ion battery cells have emerged as the linchpin, particularly in the face of the massive electrification efforts in the automotive sector driven by climate change concerns. Manufacturers are fervently establishing gigafactories to curtail production costs, thereby reshaping the industrial landscape. Central to this paradigm shift is the urgent need to optimize these batteries efficiently, as it pertains to enhancing production volume and maintaining impeccable cell quality.

This thesis project proposes a novel approach: the development of surrogate models geared towards predicting and optimizing the heterogeneities inherent in battery electrode operation. Leveraging cutting-edge artificial intelligence techniques, such as generative adversarial networks, these models aim to efficiently forecast operational variances linked to electrode conditions. By seamlessly integrating these models into optimization loops, the study seeks to offer tailored recommendations for enhancing electrode performance. Furthermore, an empirical validation phase involving electrochemical characterizations to substantiate the accuracy of the predictions is envisioned. This endeavor represents a significant stride towards revolutionizing the industrial landscape and ensuring the sustainable advancement of lithium-ion battery technology in the era of energy transformation.