Chi-Jyun KoKUO-CHING CHENCHIH-HUNG CHEN2025-05-062025-05-062025-05-01https://www.scopus.com/record/display.uri?eid=2-s2.0-105000383938&origin=resultslisthttps://scholars.lib.ntu.edu.tw/handle/123456789/728912Although constant voltage (CV) charging was traditionally used to maintain high charge levels in electrical devices, it is now proven to be an effective and rapid tool for evaluating battery states across all voltage levels. The current-based curve obtained during CV charging, called the CV curve, notably exhibits unique characteristics absent in voltage-based curves, thereby highlighting the advantages and superiority of CV charging. We find that, under the same ambient temperature and battery health state, the trajectory of CV curve is mainly determined by the voltage level during CV charging, irrelevant of the battery's initial state prior to charging and the magnitude of the initial current. The underlying mechanism of this finding is also explored. These findings, along with the observed CV curve variations across battery states, demonstrate that CV charging at various voltage levels is well-suited for the state of charge (SOC) and state of health (SOH) estimation. This study employs a small segment of the CV curve's evolution to rapidly evaluate the both battery states from the Gaussian process regression models. Analysis shows that 20-s CV data samples, at any voltage level, achieve 1 % SOC and 2.5 % SOH prediction errors, demonstrating CV curves' effectiveness for battery state estimation.Constant voltage chargingLithium-ion batteryMachine learningState of chargeState of health[SDGs]SDG7[SDGs]SDG11journal article10.1016/j.energy.2025.1357172-s2.0-105000383938