State of Charge Estimation of Lithium Sulfur Batteries using Sliding Mode Observer

The lithium-sulfur (Li-S) batteries are high energy storage systems that can be used for electric grid and solar power air vehicle applications. Such applications require an accurate state of charge (SOC) estimator to control and optimize battery performance. Modelling and estimation of discharging Li-S are highly challenging than other batteries as the discharge voltage of Li-S batteries has highly nonlinear and typical characteristics than the Lithium-Ion batteries. For Li-S battery SOC estimation, literature has proposed filters and machine learning techniques, but no literature on sliding mode observer (SMO). This paper presents the SMO for discharging Li-S SOC estimation and compares it to the extended Kalman filter (EKF). Both estimators use a first-order equivalent circuit network (ECN) model of Li-S cell parameters given in the literature. The performance of such ECN model based SOC estimators influenced by the Q- uncertainty, which is a perturbation in the form of process noise state-space model. Therefore, this work studies an optimal trade-off characteristic of SMO and EKF over the Q-uncertainty. With constant and mixed-amplitude pulse load current sequences, numerical simulation has performed. Simulation results illustrate that the SMO is optimal, converges to the true SOC than the EKF when the perturbation increased.