AI-Powered Lithium-Ion Battery State of Charge Estimation for Enhanced Electric Vehicle Efficiency

Efficient utilization of lithium-ion (Li-ion) batteries is critical for enhancing the overall performance of electric vehicles (EVs). Accurate estimation of the State of Charge (SoC) of these batteries plays a pivotal role in optimizing their operation. This study investigates the application of artificial intelligence techniques, specifically artificial neural networks (ANN), support vector machines (SVM), linear regression (LR), and K-Nearest Neighbors (KNN), for the estimation of battery SoC. The dataset employed in this research originates from Panasonic's 18650PF dataset, encompassing critical battery parameters. To facilitate accurate model predictions, a comprehensive preprocessing phase was conducted, followed by feature extraction using sparse principal component analysis (SPCA). By employing the extracted features, the AI models were trained to estimate the battery SoC. To evaluate the performance of these models, multiple evaluation metrics were adopted, including Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and the coefficient of determination (R2). Results from the comparative analysis indicated that the ANN model outperformed other AI techniques. The R2 value, a crucial indicator of model performance, was observed to be 0.9925 for the ANN model. This high R2 value signifies the ANN's ability to effectively capture the underlying patterns within the battery dataset, resulting in remarkably precise SoC estimates.