State of Charge Estimation for Lithium-Ion Batteries in Hybrid Vessels Using Kalman Filters

Abstract

Lithium-ion (Li-ion) batteries have gained significant attention in applications such as electric vehicles and hybrid vehicles and vessels. An important variable of the battery pack is the state-of-charge (SoC), which is not deterministic and cannot be measured accurately using the currently available methods, but it requires estimation. The Kalman filter algorithm has proven to be reliable in estimating the charging level of Li-ion batteries. In this paper, we compare the performance of two different Kalman-based algorithms in terms of robustness and accuracy, also distinct in the nonlinearity scheme. A polynomial curve fitting method is customized to estimate the essential Li-ion battery parameters based on its Thevenin equivalent circuit model. Then, the output of these parameter estimations is propagated through two Kalman filters: extended Kalman filter (EKF) and the unscented Kalman filter (UKF). The implemented filter algorithms are validated using realistic data collected from a functional hybrid vessel operating between the Nordic countries. Further validation for the battery parameters is done using laboratory experiments on battery packs. The SoC estimation results show that both filters EKF and UKF converge robustly and reliably. However, UKF is further recommended as it outperforms EKF in terms of battery SoC estimation accuracy with an advantage of 0.05% error, hence, making UKF more suitable for real-time Li-ion battery-based applications.