State of health (SOH) estimation is insightful for the lithium-ion battery (LIB) health management. This paper proposes a new set of health indicators (HIs) based on early-stage constant-voltage (CV) charging, which are easily available in practical vehicle applications. Particularly, a thorough analysis is performed over different CV-based HIs to obtain the informative ones with strong correlation against the SOH. A gaussian process regression (GPR) model is further employed to fusion the extracted HIs and to estimate the battery SOH. The proposed method is validated based on cycling experiments performed on the LiNiCoAlO2 cells. Results suggest that the proposed method promises multifold benefits, including the high estimation accuracy, low requirement on the charging integrity, and the high robustness to cell inconsistency.
Keywords lithium-ion battery, health indicators, state of health, constant voltage charging, Gaussian process regression