Accurate State of Charge Estimation of Lithium-Ion cell using Unscented Kalman Filter

Accurate State of Charge (SoC) estimation for Lithium-ion (Li-ion) batteries has a crucial role in ensuring the safety, extending the battery lifespan and ultimately contributes in extending the range of Electric Vehicles (EVs). Several Kalman Filter based algorithms have been widely explored in existing literature. The traditional implementations encounter certain challenges such as use of poor battery model, sluggish convergence rate due to higher complexity in algorithm and larger overshoot, slow system and observation noise covariance update, etc. These limitations may affect their capability to effectively capture the dynamic nature of system uncertainties, restricting their effectiveness in estimating SoC accurately. This paper includes Unscented Kalman Filter (UKF) algorithm for accurate estimation of SoC that uses a precise Thevenin 2RC equivalent circuit model of a li-ion cell incorporating the effect of temperature (T) and assuming higher measurement uncertainties. The elements of the measurement noise covariance (R) and the process noise covariance (Q) are tuned for faster convergence of algorithm with increased accuracy in SoC estimation. The robustness of UKF algorithm has been tested with Worldwide Harmonized Light Vehicles Test Procedure (WLTP) discharge profile under different operating conditions. Simulation was performed in Simulink-MATLAB. The Root Mean Square Error observed in estimation results was less than 0.6% which is notably lower compared to few analogous algorithms proposed in past literature. The error analysis findings indicate that the algorithm exhibits superior performance under dynamic load conditions.