Sign in Subscribe
Machine learning

Novel Machine Learning Approach Predicts Lithium-ion Battery Discharge Capacity

Strategic Insights

— 2 min read

A new study has successfully developed an interpretable machine-learning approach that accurately predicts lithium-ion battery discharge capacity under variable operating conditions using minimal input descriptors. Funded by the Interdisciplinary Research Center for Construction & Building Materials (IRC-CBM), King Fahd University of Petroleum & Minerals, the research used three nature-inspired hybrid machine learning algorithms: Quantum-inspired Particle Swarm Optimization-Adaboost (QIPSO-ADB), Harris Hawks Optimization- Extreme Gradient Boosting Machine (HHO-GBM), and Sparrow Search Algorithm-Light Gradient Boosting Machine (SSA-LGBM). The study demonstrated that discharge capacity across cycles can be accurately predicted using only temperature and cycle number, simplifying model inputs while maintaining high accuracy.

Key Takeaways:

  • The researchers developed three nature-inspired hybrid machine learning algorithms: QIPSO-ADB, HHO-GBM, and SSA-LGBM to predict lithium-ion battery discharge capacity.
  • The HHO-GBM model achieved the highest prediction accuracy with a mean absolute error (MAE) of 0.0816 and a correlation coefficient of 95.2 % during the testing phase.
  • The study demonstrated that discharge capacity across cycles can be accurately predicted using only temperature and cycle number, reducing model inputs while maintaining high accuracy.
  • The HHO-GBM-C2 model achieved a low MAE of 0.1438 and a correlation coefficient of 85.99 % when using only temperature and cycle number as input descriptors.
  • The research concluded that these findings provide strategic insights for optimizing battery performance, contributing significantly to the development of reliable electric vehicles and sustainable energy storage solutions.
  • The study used multiple samples from eVTOL and MIT public datasets to validate the robustness and generalizability of the models across both variable and fixed operating conditions.
  • The researchers involved in the study include Yakubu Sani Wudil, M. A. Gondal, and Mohammed A. Al-Osta from King Fahd University of Petroleum and Minerals.

Statistics:

  • Mean absolute error (MAE) of 0.0816 and a correlation coefficient of 95.2 % for the HHO-GBM-C1 model during the testing phase.
  • Mean absolute error (MAE) of 0.1438 and a correlation coefficient of 85.99 % for the HHO-GBM-C2 model when using only temperature and cycle number as input descriptors.
  • Prediction accuracy of 95.2 % for the HHO-GBM-C1 model and 85.99 % for the HHO-GBM-C2 model.

Sources:

  • "Investigating Lithium-ion Battery Discharge Capacity Under Variable Operating Conditions Using Nature-inspired Hybrid Algorithms With Minimal Descriptors." Journal of Energy Storage, 2025;118.
  • King Fahd University of Petroleum and Minerals.
  • Interdisciplinary Research Center for Construction & Building Materials (IRC-CBM).
  • eVTOL and MIT public datasets.

Read more