Use of supervised learning algorithms to determine the state of charge (SoC) accurately in LiFePO ₄ batteries

This study aimed to discern the optimal machine learning algorithm for precise state-of-charge (SoC) prediction in lithium-ion batteries, specifically focusing on LiFePO 4 . The investigation involved training and evaluating a comprehensive dataset using diverse machine learning algorithms, including Linear Regression (LR), Support Vector Machine (SVM), Decision Tree (DT), K-Nearest Neighbor (KNN), Artificial Neural Network (ANN), and Stochastic Gradient Descent (SGD). The results indicate that the Decision K-nearest neighbor, and Artificial Neural Network algorithms yield superior performance in predicting the SoC of Lithium-Ion batteries. Notably, the Decision Tree algorithm demonstrated the lowest error value of 7 × 10 −9 , followed by the Nearest Neighbor algorithm with an error value of 1.9 × 10 −4 , and the ANN algorithm with an error value of 0.01. These findings highlight the efficacy of machine learning algorithms in accurately predicting the SoC of Lithium-Ion batteries, which has significant implications for optimizing battery performance and efficiency. Furthermore, this study contributes to the advancement of novel and advanced machine learning algorithms for precise SoC prediction in LiFePO 4 batteries.