Scholarlink

In past years, demand for electrochemical energy storage systems has increased significantly with growth of renewable energy sector and electric vehicle (EV) market. Cathode materials in a lithium-ion battery are critical to determine its voltage and capacity output. Research community has been working to develop high-voltage and electrochemical stable cathodes via experimental and computational methods. Alternatively, machine learning models can be used that exhibit inherent advantages over the DFT in terms of computation time, less hardware intensive, and cost. In this study, artificial neural network (ANN) based machine learning model is trained to predict the voltage of lithium-ion battery cathodes. The training and test database are constructed from Materials Project (MP) database by selecting stability (energy above hull) and lithium working ion as filters. The ANN model exhibits R2 up to 0.97, MAE up to 0.11, MSE up to 0.04, and RMSE up to 0.20. The results can be useful for screening and designing cathode materials for high-voltage lithium-ion batteries.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Voltage Prediction of Lithium-Ion Battery Cathode Using Machine Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Arnav Pareek,
  • Jayesh Sharma,
  • Kartik Kumar,
  • Ramdutt Arya,
  • Kapil Pareek

摘要

In past years, demand for electrochemical energy storage systems has increased significantly with growth of renewable energy sector and electric vehicle (EV) market. Cathode materials in a lithium-ion battery are critical to determine its voltage and capacity output. Research community has been working to develop high-voltage and electrochemical stable cathodes via experimental and computational methods. Alternatively, machine learning models can be used that exhibit inherent advantages over the DFT in terms of computation time, less hardware intensive, and cost. In this study, artificial neural network (ANN) based machine learning model is trained to predict the voltage of lithium-ion battery cathodes. The training and test database are constructed from Materials Project (MP) database by selecting stability (energy above hull) and lithium working ion as filters. The ANN model exhibits R2 up to 0.97, MAE up to 0.11, MSE up to 0.04, and RMSE up to 0.20. The results can be useful for screening and designing cathode materials for high-voltage lithium-ion batteries.