State-of-Health Estimation in NCA Lithium-Ion Batteries Using Incremental Capacity Analysis and Gaussian Process Regression While Considering Capacity Recovery

Accurate state-of-health (SOH) estimation is critical for the safe and efficient use of lithium-ion batteries, particularly in electric vehicles and energy storage systems. Conventional SOH estimation methods often neglect the capacity recovery effect, which can reduce SOH prediction accuracy after storage. In this study, a hybrid method integrating incremental capacity analysis with Gaussian process regression (GPR) to account for the capacity recovery phenomenon in lithium nickel cobalt aluminum oxide cells is proposed. Health features were extracted from incremental capacity curve peaks obtained under different storage state-of-charge conditions (0-100%) at 40 °C. The proposed model achieved strong predictive performance, with mean absolute error (MAE) below 0.12% and root mean square error (RMSE) below 0.18% across all test conditions. Comparative analysis against a multilayer perceptron demonstrated the GPR-based method attained comparable error metrics while significantly reducing computational time. These findings highlight the importance of modeling capacity recovery in SOH estimation and demonstrate the potential of GPR-based approaches for reliable and computationally efficient battery management in real-world applications.