https://github.com/christianparsons98/lithium_battery_eis_classification_and_regression_2024
Files associated with the Paper: Performance Classifcation and Remaining Useful Life Prediction of Lithium Batteries Using Machine Learning and Early Cycle Electrochemical Impedance Spectroscopy Measurements by Christian Parsons, Adil Amin, and Prasenjit Guptasarma
https://github.com/christianparsons98/lithium_battery_eis_classification_and_regression_2024
Science Score: 36.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
✓codemeta.json file
Found codemeta.json file -
✓.zenodo.json file
Found .zenodo.json file -
○DOI references
-
✓Academic publication links
Links to: arxiv.org -
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (11.2%) to scientific vocabulary
Repository
Files associated with the Paper: Performance Classifcation and Remaining Useful Life Prediction of Lithium Batteries Using Machine Learning and Early Cycle Electrochemical Impedance Spectroscopy Measurements by Christian Parsons, Adil Amin, and Prasenjit Guptasarma
Basic Info
- Host: GitHub
- Owner: ChristianParsons98
- Language: Jupyter Notebook
- Default Branch: main
- Size: 144 MB
Statistics
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
Lithium Battery EIS Classification and Regression (2024)
This repository supports the research article:
Performance Classification and Remaining Useful Life Prediction of Lithium Batteries Using Machine Learning and Early Cycle Electrochemical Impedance Spectroscopy Measurements
by Christian Parsons, Adil Amin, and Prasenjit Guptasarma
View on arXiv
Description
This code implements a lightweight, interpretable pipeline for: - Classifying lithium-ion batteries into high- and low-performing groups. - Predicting the remaining useful life (RUL) early in the battery's cycle life using only a small subset of EIS measurements.
The standout contribution is that classification and regression are performed using just 12 EIS frequencies and temperature, demonstrating 100% classification accuracy and R > 0.96 for RUL predictionwithin the first 20 cycles.
Techniques Used
- Support Vector Machine (SVM): For high/low performance classification using the 20 kHz impedance feature.
- Lasso Regression: For RUL prediction from -Im(Z) at 20 kHz and 8.8 Hz + temperature.
- Principal Component Analysis (PCA): Used in exploratory data visualization.
- Comparison with Decision Trees and Random Forests to highlight the SVMs generalization capabilities.
Libraries and Tools
Project Structure
plaintext
.
Cleaned_Data/
DE_Figures/
Data/
Figures/
Classification_Low_Cycles_*.ipynb
Regression_Low_Cycles_Cleaned_*.ipynb
Data_Cleaning.ipynb
PCA_EIS_Exploration.ipynb
README.md
Notable Directories
Data/: Raw Zhang dataset and EIS measurements.Cleaned_Data/: Aligned and processed EIS + RUL data.Figures/,DE_Figures/: Generated figures used in paper and for presentation.*_Low_Cycles_*.ipynb: Notebooks analyzing each SOC condition (III, V, IX).
Highlights from the Paper
- Single-feature SVM classification using 20 kHz impedance achieves 100% accuracy.
- RUL prediction using only impedance at 20 kHz and 8.8 Hz + temperature yields R > 0.96.
- Models are robust across stable and unstable SOC states.
- No complex preprocessing or high-dimensional ML is neededresults are reproducible and interpretable.
Citation
If you use this code or build upon this work, please cite:
arXiv:2408.03469
Owner
- Login: ChristianParsons98
- Kind: user
- Repositories: 1
- Profile: https://github.com/ChristianParsons98
GitHub Events
Total
- Watch event: 1
- Push event: 1
Last Year
- Watch event: 1
- Push event: 1