Data-driven health-aware battery fast charging using adaptive neuro-fuzzy inference system-based optimized end-charging strategy for battery life extension

Fast charging techniques have attracted significant attention for their potential to substantially reduce charging time, despite the associated risk of accelerated degradation. However, most existing fast-charging strategies rely on open-loop control, limiting their ability to mitigate degradation across varying operating conditions. In response to this challenge, the study introduces a data-driven, health-aware battery fast charging (HABFC) approach that employs an adaptive neuro-fuzzy inference system (ANFIS) to determine the real-time transition point between constant current (CC) and constant voltage (CV) operation. The proposed framework integrates knowledge-driven and data-driven paradigms, where expert-defined rule-based guidelines provide structural initialization and enforce safety constraints, while the ANFIS model captures complex nonlinear relationships between battery states and optimal charging decisions. This hybrid approach enables accurate and adaptive prediction of the CC-CV transition, ensuring both operational safety and physical consistency across diverse conditions. To enhance robustness, the strategy incorporates state-of-health (SoH) and state-of-charge (SoC) as key inputs. A real-time degrading equivalent circuit model (ECM) is employed to estimate SoC, thereby mitigating estimation errors and improving decision reliability. The resulting ANFIS-based HABFC strategy facilitates dynamic, real-time switching between CC and CV modes, achieving faster charging without compromising battery longevity. The proposed method is experimentally validated using a customized charging setup. Results demonstrate approximately 12% reduction in charging time alongside a 23% improvement in cycle life compared to the conventional charging technique. The electrochemical impedance study indicates mitigation of lithium plating by reduced impedance growth, which stabilizes electrode kinetics, thereby enhancing electrochemical stability and prolonging battery life. Furthermore, comprehensive comparisons with benchmark strategies highlight superior performance across charging efficiency, degradation mitigation, and thermal behavior. Overall, this work establishes the effectiveness of real-time adaptive CC-CV switching at the cell level and provides a scalable foundation for future extension to battery pack-level applications.