Work

Scientific Machine Learning for Battery Research

Year: 2025

Type: article

Abstract: This talk introduces key Machine Learning concepts with applications to battery research: Traditional Machine Learning: example applying symbolic regression to predict conductivity of Li‑ion battery electrolytes. Neural Networks: applications for predicting degradation trajectories of fast‑charged batteries. Explainable Artificial Intelligence: methods to interpret how neural networks classify infrared spectra for presence/absence of functional groups. Physics‑Informed Neural Networks (PINNs): hybridizing phsysical models and neural networks, with applications in materials science and electrochemistry. We argue for a critical assessment of AI beyond the hype: understanding AI’s strengths and limitations is essential to identify value‑generating applications that truly accelerate battery research.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.17193294

Author: Flores, Eibar

Institution: SINTEF Industry

Language: English

Topic: Machine Learning in Materials Science

Subfield: Materials Chemistry

Field: Materials Science

Domain: Physical Sciences

SDG: Affordable and clean energy

Open Access status: green

Funders: European Commission, Senter for Internasjonalisering av Utdanning