High Performance Battery SoC Estimation Method based on an Adaptive Square-Root Unscented Kalman Filter

Nowadays, high performances and reliability are being required from the energy storage system in different fields of application, including hybrid and electric vehicles as well as grid-tied power converters for the integration of renewable energy sources. Therefore, with the increasing of lithium-ion batteries (LIBs) application, the estimation of the State of Charge (SoC) at both cell and pack levels represents a fundamental task to be performed by the Battery Management System (BMS) in order to optimally manage the operating conditions, maximize the overall performances while extending the useful life. However, the nonlinear characteristics of the LIB parameters and the measurement noise due to the accuracy of current and voltage sensors in real applications strongly affect the performances of traditional SoC estimation methods, such as Coulomb counting and Open Circuit Voltage observation, leading to significant errors. To overcome these issues, different model-based methods have been proposed in literature. In this paper, Kalman filter methods are investigated with aim of evaluating the variability of their performances with respect to the initial calibration of the covariance matrices and the specific operating condition. In addition, the benefits of a proposed adaptive algorithm combined with a state estimator are highlighted.