Observational analysis demonstrates high accuracy in battery health and capacity estimation in electric vehicles, suggesting a novel efficient framework.
Key Points
Battery health anomaly detection achieved an accuracy of 98.9%, indicating superior performance over existing approaches.
Mean absolute error in capacity estimation was 0.32 Ah, corresponding to a relative error of approximately 1.80%.
The Split Multi-Task Federated Learning framework effectively combines the advantages of split learning and federated learning.
Energy consumption is minimized by performing computations during charging sessions, enabling practical real-world applications.