Estimation of state of charge for a lithium-ion battery pack using nonlinear Kalman filters

Estimating the state of charge (SOC) of a lithium-ion battery (LiB) pack is challenging due to the inherent variability across individual battery cells. This study uses a hardware configuration comprising a 13s10p battery pack, a switched-mode power supply (SMPS), a brushless direct current motor (BLDC) as a load, and a charger to charge and discharge the battery pack for gathering the real-time data. The data is subsequently fed into the simulation model, which estimate the SOC for a 2 RC model at temperatures 288 K, 298 K, and 318 K. Several nonlinear Kalman filter (KF) techniques, such as the extended Kalman filter method (EKF), the unscented Kalman filter method (UKF), extended Kalman-Bucy filter method (EKBF), and the unscented Kalman-Bucy filter method (UKBF), are utilized in estimating SOC. The UKBF and EKBF provide the most accurate estimation for SOC, with an overall root mean square error (RMSE) of less than 1% and 1.5%, respectively, while the mean absolute percentage error (MAPE) is below 1.5% and 3% for the 2 RC model across all temperatures.