Integrated framework for battery cell state-of-health estimation in complex modules: Combining current distribution analysis and novel terminal voltage estimation L-EKF modeling

In electric vehicles, the variability among individual cells within power battery modules presents formidable obstacles in determining the state-of-health (SOH). This study reveals an innovative approach for estimating the terminal voltage of battery cells in modules, with a focus on enhancing the accuracy of SOH estimation for individual cells. First, the topological structure of the current distribution within the battery module is thoroughly analyzed to establish a probabilistic model of the current distribution. Subsequently, a hybrid algorithm combining a long short-term memory (LSTM) network and an extended Kalman filter (EKF), referred to as the L-EKF, is proposed to realize terminal voltage estimation. Additionally, the use of an adversarial domain distribution probability network (AdaDDPN) further augments the accuracy of SOH estimation for individual cells. The experimental results demonstrate that the proposed method outperforms alternatives in terms of error metrics, including the MAE, RMSE, and MAXE. The battery cell terminal voltage estimation exhibited an RMSE of 6.31 mV and an MAXE of 25.01 mV, with AdaDDPN achieving MAEs below 0.02 and RMSEs below 0.025. This approach not only provides a reliable method for assessing individual cell health in battery management systems but also significantly enhances the safety and reliability of electric vehicles.