Optimized estimation of Li-ion battery parameters to improve the Enhanced Self-Correcting model with nonlinear least-squares data fitting and SoC estimation using Invariant EKF

In recent years, batteries have become increasingly important, especially lithium batteries. Battery modeling is essential for many applications, such as state-of-charge (SoC) estimation. To obtain a high-performance model, the estimation of battery parameters must be precise. The Enhanced Self-Correcting model describes several battery dynamics, such as hysteresis cycling. This model can be further improved by making these parameters variable as a function of the SoC and temperature, making the model more efficient. In this study, the parameters of the single RC branch circuit during charging and discharging will be estimated, and these parameters will then be used to improve the “Enhanced Self-Correcting” model, which will enable it to better describe battery dynamics and improve accuracy. MATLAB/Simulink Toolboxes can simplify many tasks, and nonlinear least-squares data fitting using the Trust-Region-Reflective algorithm produces remarkable results. Then, this model is validated with the “SoC and voltage” estimation compared to other models, using the invariant extended Kalman filter (IEKF), which is reliable for nonlinear systems as its correction is independent of the output error, leading to greater accuracy and performance.