Combined Capacity and Internal Resistance Estimation of Lithium-Ion Batteries Based on Voltage Recovery

Accurate estimation of a battery's aging state is a critical prerequisite for ensuring the safety of electric vehicles. Currently, the prevailing data-driven methods for estimating battery states prioritize capacity estimation, often disregarding internal resistance. This focus may precipitate safety risks arising from battery inconsistencies. Consequently, it is essential to accurately estimate both the capacity and internal resistance of batteries in tandem to ensure reliable and consistent battery testing. Hence, a novel method leveraging the WOA-ELMAN neural network for the simultaneous estimation of battery capacity and internal resistance has been proposed to tackle the aforementioned challenge. This approach identifies voltage recovery as a key health indicator, gleaned from the historical performance data of the battery. Subsequently, an integrated model for estimating both capacity and internal resistance is developed using the ELMAN neural network architecture. The whale optimization algorithm is then engaged to fine-tune the initial parameters, thereby enhancing the estimation accuracy. To validate the efficacy of this method, the “Path Dependent Battery Degradation Dataset” from Oxford-a widely recognized public dataset-is employed. The findings reveal that the average root mean square error (RMSE) for capacity and internal resistance estimation stands at 0.89% and 7.14%, respectively.