Abstract

The accuracy of single particle (SP) models for lithium-ion batteries at high C-rates is constrained by lithium concentration gradients in the electrolyte, which affect ionic conductivity, overpotential, and reaction rates. This study addresses these limitations using extreme gradient boosting machine learning (ML). By training our ML model with data from a comprehensive electrochemical (P2D) model and performing sensitivity analysis on key battery parameters, we enhance predictive accuracy. Compared to conventional SP and P2D models under constant current loading, our ML-based SP model achieves similar predictive accuracy to P2D, with significant improvements in computational efficiency. Additionally, the ML-based SP model demonstrates improved predictive accuracy under dynamic loading conditions, providing a practical framework for improving battery management and safety.

Department(s)

Mechanical and Aerospace Engineering

Publication Status

Available Access

Comments

National Science Foundation, Grant None

International Standard Serial Number (ISSN)

0003-6951

Document Type

Article - Journal

Document Version

Final Version

File Type

text

Language(s)

English

Rights

© 2024 American Institute of Physics, All rights reserved.

Publication Date

30 Sep 2024

Share

 
COinS