State-of-Health Estimation and Remaining Useful Life Prediction of Lithium-Ion Batteries Based on Optimized Echo State Network With Early Data Feature

Accurate estimation and prediction of the state-of-health (SOH) and remaining useful life (RUL) of lithium-ion batteries (LIBs) at the early cycling stage are essential for enabling efficient battery recycling, secondary utilization, and timely warnings. However, the high similarity of battery charging data during the early cycling stage, the requirement for extensive data to assess aging, and the complexity of multidimensional feature extraction pose significant challenges for existing predictive methods. To address these challenges, a novel framework based on an optimized echo state network (ESN) for SOH estimation and RUL prediction with early data features is proposed. Inspired by the idea of phase-space reconstruction, the delay time is employed to capture the characteristics of the early voltage profile. A novel improved whale optimization algorithm (IWOA) is employed to optimize the ESN, facilitating rapid and precise prediction of both SOH and RUL. Experimental results showed that the root mean square error (RMSE) and mean absolute percentage error (MAPE) for battery SOH can be reduced to 0.33% and 0.27%, respectively, and the RMSE and MAPE for battery RUL can be reduced to 0.4% and 0.43%, respectively. By leveraging early cycle data, the proposed method not only enhances the efficiency and accuracy of SOH estimation and RUL prediction, but also introduces a novel perspective for practical battery management and predictive maintenance, thereby advancing the state-of-the-art in battery health monitoring systems.