Answers are grounded in this paper and cite their sources.
As electric vehicles (EVs) reach the end of their primary lifespan, the number of retired EV batteries continues to grow rapidly. These batteries can be repurposed for second-life stationary applications, such as energy storage systems, because they still retain sufficient capacity for less demanding use cases. Effective and accurate battery screening is essential for reliable system integration, owing to differences in the usage histories and operating conditions of retired EV batteries. Accurately estimating state of health (SOH), which is quantified from the remaining capacity, is therefore critical. Traditionally, SOH estimation relies on full charge–discharge test protocols, which are reliable but time-consuming, making them impractical for rapid screening of large numbers of retired EV batteries. This study investigates 12 health features (HFs) using 466 commercial cells from Nissan Leaf EV batteries, with SOH ranging from approximately 50% to 70%. The HFs are categorized into two groups. The first group includes the proposed features extracted from partial constant-current (CC) charging without requiring initial state-of-charge (SOC) control. The second group includes the conventional features obtained from constant-voltage charging, voltage relaxation, and incremental capacity analysis. To enable the rapid screening and sorting of retired EV batteries, the extracted HFs were evaluated using three models of varying complexity: linear regression, Gaussian process regression, and multilayer perceptron. The results show that the proposed HFs obtained from partial CC charging achieve estimation errors comparable to those of conventional HFs, suggesting that reliable sorting is feasible without controlling the initial SOC.
No takes yet. Share an insight, caveat, or question.
Lyu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: