SOC-SOH co-estimation across the battery life cycle using a simplified electrochemical model with adaptive parameter updating and LSTM

As a core function of the battery management system, state of charge (SOC) estimation has been the focus of investigation. However, previous research does not adequately consider the bidirectional interference effect between battery state of health (SOH) and SOC estimation. This leads to an increase in the error in SOC estimation after battery aging. The role of bidirectional interference is addressed by a hybrid algorithmic framework that updates the battery model parameters and co-estimates SOC/SOH. Specifically, a simplified electrochemical model (SEM) with online parameter update is first constructed in this paper, and the SOC estimation algorithm is built based on this model with the EKF algorithm. The SEM parameters are updated with the help of parameter sensitivity analysis. Subsequently, a battery SOH estimation model is built using the long short-term memory (LSTM) algorithm. It should be noted that the LSTM algorithm uses voltage and current data from the battery when its SOC is between 20% and 80%. Finally, a hybrid SOC/SOH estimation framework with multi-time scale update is constructed. Based on the method proposed in this paper, the SOC estimation error under the New European Driving Cycle (NEDC) working condition throughout the life cycle of the lithium-ion battery is less than 0.8%.