| Autor: |
Li, Hongbo, Li, Zebin, Ma, Yongchun, Lin, Jie, Zhao, Xiaobin, Zhang, Wencan, Guo, Fang |
| Předmět: |
|
| Zdroj: |
AIP Advances; Jul2024, Vol. 14 Issue 7, p1-11, 11p |
| Abstrakt: |
Energy storage batteries still have usable capacity after retirement, with excellent secondary utilization value. Estimating the state of health (SOH) of retired batteries is critical to ensure their reuse. As the battery first reaches the end of its useful life, its performance degradation pattern significantly differs from that in service, increasing the difficulty of accurate SOH estimation. This study developed a SOH estimation method for retired batteries based on battery positive, negative, and center temperature data from 80% to 50% of retired battery health. The variational mode decomposition technique divides the temperature signal into multiple trends representing different battery aging mechanisms. The decomposed modes are given a physical meaningfulness, providing a new perspective to monitor battery health. In addition, this study proposes a multi-task learning framework that realizes the parallel processing of two tasks under this framework. On the one hand, the gated recurrent unit is used to estimate the relationship between the battery baseline temperature and SOH, which captures macro-degradation trends of the battery. On the other hand, the transformer network is responsible for analyzing short-term battery health fluctuations caused by subtle temperature changes. This multi-task approach can simultaneously process and analyze both macro-degradation trends and micro-fluctuations in battery degradation, estimating that the root mean square error of battery health is 5.22 × 10−5. Compared to the existing techniques, this study shows potential applications in the retired battery state of health assessment. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
