Random Forest-Based Machine Learning Model Design for 21,700/5 Ah Lithium Cell Health Prediction Using Experimental Data DOI Creative Commons
Sid-Ali Amamra

Physchem, Journal Year: 2025, Volume and Issue: 5(1), P. 12 - 12

Published: March 16, 2025

In this research, the use of machine learning techniques for predicting state health (SoH) 5 Ah—21,700 lithium-ion cells were explored; data from an experimental aging test used to build prediction model. The main objective work is develop a robust model battery estimation, which crucial enhancing lifespan and performance batteries in different applications, such as electric vehicles energy storage systems. Two models: support vector regression (SVR) random forest (RF) designed evaluated. model, novel strategy SoH application, was trained using features, including current (A), potential (V), temperature (°C), tuned through grid search optimization. developed models evaluated two metrics, R2 root mean squared error (RMSE). obtained results show that outperformed SVR achieving 0.92 RMSE 0.06, compared 0.85 0.08 SVR. These findings demonstrate effective prediction, offering promising alternative existing monitoring strategies.

Language: Английский

Random Forest-Based Machine Learning Model Design for 21,700/5 Ah Lithium Cell Health Prediction Using Experimental Data DOI Creative Commons
Sid-Ali Amamra

Physchem, Journal Year: 2025, Volume and Issue: 5(1), P. 12 - 12

Published: March 16, 2025

In this research, the use of machine learning techniques for predicting state health (SoH) 5 Ah—21,700 lithium-ion cells were explored; data from an experimental aging test used to build prediction model. The main objective work is develop a robust model battery estimation, which crucial enhancing lifespan and performance batteries in different applications, such as electric vehicles energy storage systems. Two models: support vector regression (SVR) random forest (RF) designed evaluated. model, novel strategy SoH application, was trained using features, including current (A), potential (V), temperature (°C), tuned through grid search optimization. developed models evaluated two metrics, R2 root mean squared error (RMSE). obtained results show that outperformed SVR achieving 0.92 RMSE 0.06, compared 0.85 0.08 SVR. These findings demonstrate effective prediction, offering promising alternative existing monitoring strategies.

Language: Английский

Citations

1