Calendar ageing modelling using machine learning: an experimental investigation on lithium ion battery chemistries

Area of Science:

• Materials Science
• Electrochemistry
• Data Science

Background:

• Calendar ageing significantly impacts battery performance and operational lifespan.
• Predicting capacity degradation is crucial for identifying and mitigating battery ageing issues.
• Understanding ageing mechanisms is vital for enhancing battery longevity in various applications.

Purpose of the Study:

• To evaluate the efficacy of machine learning algorithms for predicting battery calendar ageing.
• To compare the predictive performance of Extreme Gradient Boosting (XGBoost) and artificial neural networks (ANN) on diverse battery chemistries.
• To identify the most suitable algorithm for electric vehicle battery calendar ageing prediction.

Main Methods:

• Utilized machine learning algorithms: Extreme Gradient Boosting (XGBoost) and Artificial Neural Network (ANN).
• Applied algorithms to predict calendar ageing data across six cell chemistries: LCO, LFP, LMO, LTO, NCA, and NMC.
• Evaluated model performance using Mean Absolute Percentage Error (MAPE) and Mean Absolute Error (MAE).

Main Results:

• XGBoost achieved an overall MAPE of 0.0126, outperforming ANN's overall MAE of 0.0472.
• XGBoost demonstrated exceptional performance for Nickel Cobalt Aluminum Oxide (NCA) and Nickel Manganese Cobalt Oxide (NMC) chemistries with MAPEs of 0.0035 and 0.0057, respectively.
• ANN showed poor fitting performance for NMC at 100% state of charge and 60°C compared to XGBoost.

Conclusions:

• XGBoost is the preferred algorithm for electric vehicle battery calendar ageing prediction due to its lower error rates and superior fitting performance.
• NCA and NMC chemistries, widely used in EV batteries, showed particularly strong prediction results with XGBoost.
• The findings support the adoption of XGBoost for reliable battery health monitoring and management in electric vehicles.