A novel hybrid electrochemical equivalent circuit model for online battery management systems

Accurate battery modeling and parameter identification play pivotal roles in ensuring safety and reliability across the entire battery life cycle. Equivalent circuit models (ECM) are convenient but do not represent physical characteristics well; in contrast, electrochemical models with strong physical meaning are hard to parameterizing in an online setting. To address these challenges, this paper introduces a novel hybrid electrochemical Equivalent Circuit Model (eECM), which integrates electrochemical processes into an ECM, representing slow-dynamic internal processes with a simplified representation of solid- and liquid-phase diffusion; fast-dynamics are represented by ECM terms. The model is supported by an Adaptive Extended Kalman Filter (AEKF) to manage battery state changes and mitigate noise. To enhance parameter identification, a Fisher information matrix-enhanced Variable Forgetting Factor Recursive Least Squares (Fisher-VFFRLS) approach is employed, guided by the Cramér–Rao bound for identifying the most sensitive data points directly from the discharge cycle. Electrochemical parameters are determined via post-charging rest via a Genetic Algorithm (GA). The proposed methodology is validated on three dynamic cycles—DST, US06, and FUDS-demonstrates the effectiveness of the proposed eECM and parameter identification strategy, with maximum Root Mean Square Error (RMSE) for terminal voltage and State of Charge (SoC) estimation below 0.0076 and 0.0122, respectively.