Identification of Optimal Multivariate Factors Based on Variability Analysis of EIS Data and SOH Estimation Using AdaBoost-GRU Model

The precise estimation of the state-of-health(SOH) of lithium-ion batteries(LIBs) in electric vehicles(EVs) is crucial for the maintaining optimal performance, reliability, and safety. However, the intricate electrochemical processes within batteries and operational constraints pose significant challenges to accurate SOH estimation, particularly in real world scenarios. This research leverages electrochemical impedance spectroscopy(EIS) data to extract frequency-based parameters for integration into a battery SOH estimation model. The sensitivity of battery degradation to impedance changes over time is analyzed to identify key frequency factors that reflect aging independently of the state-of-charge(SOC) and ambient temperature, thereby minimizing external influences. An AdaBoost-GRU ensemble model is developed to utilize the derived optimal frequencies as input. A gated recurrent unit(GRU) model is embedded within the AdaBoost algorithm to prevent overfitting by adjusting and combining weights for SOH estimation. The performance of the proposed model is validated through its ability to provide accuracy with all SOH estimation errors within 2%.