Answers are grounded in this paper and cite their sources.
Assessing the technical condition of lithium-ion batteries in electric vehicles is important for the safety, performance and sustainable development of vehicles. This study introduces machine learning models to accurately and efficiently predict the technical state of lithium-ion batteries, classifying them into three categories: batteries in good condition, those with over discharge fault, and those with short circuit fault. Among the evaluated models – Fine Tree, Fine Gaussian Support Vector Machine, and Support Vector Machine Kernel – the Fine Gaussian Support Vector Machine demonstrated superior performance, achieving 95.2% accuracy in just 3.2785 sec. All three models correctly predicted the short-circuited battery with 100% accuracy. In predicting the battery being in over discharge condition, the Fine Gaussian SVM model predicted 96.7% correctly and 3.3% incorrectly with the highest accuracy, the Fine tree model predicted 82.7% correctly, with 17.3% incorrectly, the Kernel SVM model predicted with higher accuracy than the Fine tree model with 90.7% correct prediction, 9.3% incorrectly predicting the battery being in over discharge condition. Fluctuations in the input space were also propagated to investigate the robustness of the model under the presence of variability. Moreover, to facilitate rapid and precise battery condition assessment, the research team developed a predictive support tool (Graphical User Interface) that integrates the best machine learning model into a practical solution. This tool offers a convenient and effective approach for managing and maintaining lithium-ion batteries in electric vehicles, contributing to safer and more efficient operations.
No takes yet. Share an insight, caveat, or question.
Tien-Thinh Le (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: