Code for Optimized ANN-Based Prediction of Battery Capacity Using Voltage/Current Cycling Data. Related Paper: “Computational Micromechanics and Machine Learning-Informed Design of Composite Carbon Fiber-Based Structural Battery for Multifunctional Performance Prediction”
DOI: 10.4121/2040ea92-10a9-4b56-b1b0-36bcaddf0762
Datacite citation style
Software
This repository contains MATLAB code for predicting battery capacity using an Artificial Neural Network (ANN) trained on structured cycling data. The script utilizes Bayesian optimization to fine-tune hyperparameters, enabling more accurate forecasting of capacity degradation over time. This code was used in the paper titled: "Computational Micromechanics and Machine Learning-Informed Design of Composite Carbon Fiber-Based Structural Battery for Multifunctional Performance Prediction."
It is a clear and modular code that takes voltage and current data as input features, performs normalization, splits the data into training/validation/testing sets, and builds an ANN using MATLAB’s Deep Learning Toolbox. In my case, the code was applied to carbon fiber-based structural battery data to evaluate long-term electrochemical performance. This code was developed during my Master’s research at KAIST (Korea Advanced Institute of Science and Technology).
History
- 2025-05-19 first online, published, posted
Publisher
4TU.ResearchDataFormat
MATLAB/.m/.mat Image/.pngAssociated peer-reviewed publication
Computational Micromechanics and Machine Learning-Informed Design of Composite Carbon Fiber-Based Structural Battery for Multifunctional Performance PredictionOrganizations
Korea Advanced Institute of Science & Technology (KAIST), Department of Mechanical EngineeringTU Delft, Faculty of Aerospace Engineering, Department of Aerospace Structures and Materials
DATA
Files (10)
- 3,939 bytesMD5:
e042f3c5f3a1d1e7b9d8f3f90998cf05
README Text.txt - 3,939 bytesMD5:
e042f3c5f3a1d1e7b9d8f3f90998cf05
README.md - 666 bytesMD5:
720550b4cb4a56d11cdaec7513fb8e61
extract_charge_preprocessing.m - 204 bytesMD5:
1689492f1cad4d1b996b0163ea16977e
extract_discharge.m - 319,267 bytesMD5:
36d4d43312731b8f8e0da6b5989ce0b9
Framework.png - 1,072 bytesMD5:
7a842e2c4399471cc1e499916e0ff38a
LICENSE.txt - 302 bytesMD5:
f93cf7b6208214a908f46cc0e4e0933a
minmax_norm.m - 6,502 bytesMD5:
1344d1a70507fdbb8443fd73728cfe88
Optimization & ANN_BatteryCapacityEstimation.m - 5,978 bytesMD5:
259a2b9b366a0c865fc959bb17764dbd
optimize_hyperparameters.m - 46,594,813 bytesMD5:
918ca3f70b403d8904dcbec8db10ca0f
SPE40_0p1C.mat -
download all files (zip)
46,936,682 bytes unzipped