Scholarlink

<p>Accurate early prediction of battery lifetime is crucial for battery development but remains challenging due to limited data availability and weak degradation signatures at early stages. Machine learning offers a promising approach to address this limitation. This study aims to develop a reliable model for predicting battery cycle life using early aging data, based on the hypothesis that features derived from early electrical and thermal measurements are sufficient for accurate cycle life estimation. The proposed framework combines feature extraction and selection with machine learning regression models using current, capacity, temperature variation, and state-of-health indicators as inputs. The model was validated on data from the first five check-ups of 45 aged battery cells. XGBoost achieved the best performance, yielding an RMSE of 63 cycles and an R<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(^2\)</EquationSource> </InlineEquation> of 0.94. These results demonstrate that the proposed approach enables accurate early-stage battery lifetime prediction and can support improved battery health management.</p>

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

Advances in battery health management: a machine learning framework for cycle number prediction

  • 期刊论文

  • 发表日期：

    2026年7月9日

查看全文
  • Jakub Tomeš,
  • Daniela Janstová,
  • Shayestegan Mohsen,
  • Jan Mareš

摘要

Accurate early prediction of battery lifetime is crucial for battery development but remains challenging due to limited data availability and weak degradation signatures at early stages. Machine learning offers a promising approach to address this limitation. This study aims to develop a reliable model for predicting battery cycle life using early aging data, based on the hypothesis that features derived from early electrical and thermal measurements are sufficient for accurate cycle life estimation. The proposed framework combines feature extraction and selection with machine learning regression models using current, capacity, temperature variation, and state-of-health indicators as inputs. The model was validated on data from the first five check-ups of 45 aged battery cells. XGBoost achieved the best performance, yielding an RMSE of 63 cycles and an R \(^2\) of 0.94. These results demonstrate that the proposed approach enables accurate early-stage battery lifetime prediction and can support improved battery health management.