Remaining useful life prediction of lithium-ion batteries by leveraging end-of-charge characteristics

Predicting the Remaining Useful Life (RUL) of lithium-ion batteries (LIBs) is crucial for ensuring their reliable performance and extending their lifespan, especially in applications such as electric vehicles (EVs). Traditional RUL prediction methods mainly focus on battery capacity degradation, but obtaining accurate capacity data in real-world settings is often challenging, leading to limited prediction accuracy. To address this issue, our study experimentally investigates the degradation behavior of EV LIBs by analysing variations in Constant-Current (C C) and Constant-Voltage (C V) charging durations at different stages of battery ageing. The results show significant changes in the rate of current decrease during the C V charging stage and the increase in Open Circuit Voltage (OCV) during the C C charging stage, resulting in varied charging times across degradation stages. Given the challenges posed by partial charging and uncertain degradation rates in practical scenarios, we propose a novel RUL prediction method. This method utilizes the C V charging characteristic, which is less affected by partial charging, and integrates Bayesian theory with real-time monitoring data to dynamically update model parameters, enhancing prediction accuracy. Our approach offers a practical, data-driven solution for more reliable LIB RUL prediction, providing a new tool for battery health management in EV applications. • Remaining useful life of Lithium-ion battery was more accurately predicted. • Current decrease rate and open circuit voltage increase rate are studied. • It is found that C V charging characteristic is less affected by partial-charging. • Dynamically updating model parameters improves prediction accuracy.