Related Experiment Video
Updated: Sep 19, 2025
10:41
The Effect of Charging and Discharging Lithium Iron Phosphate-graphite Cells at Different Temperatures on Degradation
Published on: July 18, 2018
15.6K
High C-rate Li-NMC/graphite pouch cell end-of-life prediction via cycle-dependent variations and machine learning
Jung-Goo Choi1,2, Jethro Daniel Pascasio1,2, Jaeyoung Lee1,2,3
1Department of Environment and Energy Engineering, Gwangju Institute of Science and Technology (GIST), 123 Cheomdangwagi-Ro, Buk-gu, Gwangju 61005, Republic of Korea. jaeyoung@gist.ac.kr.
Dalton Transactions (Cambridge, England : 2003)
|June 2, 2025
Summary
Accurately predicting lithium-ion battery end-of-life (EoL) is vital. This study uses machine learning and incremental capacity analysis to forecast battery degradation, improving safety and reliability in electric vehicles.
Area of Science:
- Electrochemistry
- Materials Science
- Machine Learning
Background:
- Accurate end-of-life (EoL) prediction for lithium-ion batteries is critical for electric vehicles (EVs) and energy storage systems (ESS).
- Battery degradation under high C-rate conditions impacts safety, reliability, and cost-efficiency.
- Existing methods may not fully capture complex degradation patterns.
Purpose of the Study:
- To investigate degradation characteristics of Li-NMC/graphite pouch cells under high C-rate conditions.
- To develop and validate a machine learning-based predictive model for EoL estimation.
- To enhance battery management systems (BMS) with reliable health monitoring.
Main Methods:
- Utilized incremental capacity analysis (ICA) to extract electrochemical degradation features.
- Employed ensemble machine learning models: Random Forest, Gradient Boosting, and CatBoost.
- Trained and tested models to predict the cycle number at which State of Health (SoH) reaches 80%.
Main Results:
- The Gradient Boosting model demonstrated the highest prediction accuracy for EoL.
- Achieved a root mean squared error (RMSE) of 17.63 and a mean absolute percentage error (MAPE) of 3.11.
- Successfully predicted the cycle life with high precision.
Conclusions:
- Data-driven machine learning approaches offer a reliable method for battery health monitoring.
- The proposed framework significantly advances predictive maintenance strategies for EVs and ESS.
- Accurate EoL prediction enhances battery safety, reliability, and economic viability.
