Capacity estimation of retired lithium-ion batteries using random charging segments from massive real-world data

The repurposing of retired lithium-ion batteries from electric vehicles is a critical strategy for reducing carbon emissions. Capacity estimation plays a key role in facilitating this process. However, achieving an accurate estimation remains challenging due to the inherent variability of batteries and the complexity of real-world conditions. Existing methods often rely on small datasets and feature extraction, which limits their generalizability and leads to high testing costs. To overcome these limitations, we compile the two largest known datasets of retired lithium-ion batteries and develop a tailored neural network model capable of directly capturing both long-term and short-term temporal patterns. By utilizing randomly segmented charging curves as input, the proposed model reduces the cost of charging-discharging tests in practical applications. The experimental results demonstrate that the model achieves satisfactory performance in capacity estimation, supporting the optimization of resource use and reduction of environmental pollution.