Accurate remaining capacity prediction is of great significance to ensure battery safety and improve state of charge (SOC) estimation. However, current prediction methods neglect user charging habits, leading to difficult feature extraction and inaccurate capacity estimation. This study proposes a partial voltage-based remaining capacity estimation method considering user charging behavior. Firstly, the coverage probability of various SOC charging processes based on real-world driving data is analyzed and the charging duration at equal voltage is extracted as the health feature. Then, the correlation between the health feature and remaining capacity is calculated based on the Pearson correlation coefficient. Subsequently, a joint optimization objective function is established, combining cumulative probability density function and correlation, to determine the optimal voltage range using a particle swarm optimization algorithm. Finally, a temporal convolutional network model is proposed to capture the implicit nonlinear relationship between charging duration within the optimal voltage range and remaining capacity. Across-validation shows that the proposed can track the reference capacity accurately, and the estimation error can be guided within 2.3% of the nominal capacity.
A Partial Charging Curve-Based Remaining Capacity Prediction of Lithium-Ion Batteries for Electric Vehicles
2024-10-10
825971 byte
Conference paper
Electronic Resource
English
An adaptive remaining energy prediction approach for lithium-ion batteries in electric vehicles
Online Contents | 2016
|Model-Based Fast Charging Control of Lithium-Ion Batteries in Electric Vehicles
TIBKAT | 2024
|Electric vehicle charging remaining time prediction method and system
European Patent Office | 2023
|