Accurate estimation of lithium-ion battery state of health and capacity is critical for intelligent battery management. This study develops a multi-variable cross-condition capacity estimation model based on incremental capacity (IC) curve features. First, the IC curve area is extracted to construct a health indicator. To capture the coupled, non-linear effects of temperature and discharge current on capacity fade, a temperature-zoned modeling framework is implemented. Specifically, first-order linear polynomials are applied for room temperature conditions to prevent overfitting, while second-order polynomials with interaction terms are utilized for high and low temperature conditions to model complex degradation behaviors. Furthermore, to mitigate estimation errors caused by individual battery inconsistency and varying initial states across different operating conditions, the capacity retention rate (CRR) and health indicator retention rate metrics are defined and integrated into the estimation framework. Validation across multiple dynamic operating conditions demonstrates that the optimized CRR-based model achieves an average root mean square error of 0.0261 Ah and a mean absolute percentage error of 2.83%. The proposed temperature-zoned approach provides a robust, data-driven methodology for cross-condition battery health monitoring.
Han et al. (Wed,) studied this question.