State-of-Charge Estimation of an Experimentally Identified Lithium-ion Cell Model using Advanced Nonlinear Filters

Abstract

This paper considers the state-of-charge (SoC) estimation problem for Lithium-ion (Li-ion) cells. An accurate SoC estimation is crucial in many aspects. Firstly, it prolongs battery lifespan by preventing overcharging and overdischarging. Additionally, preventing overcharging, and hence thermal runaway, averts any potential risk of fire or explosion in the battery systems. Understanding the SoC enables the efficient utilization of a battery’s capacity, enhancing the performance of the device or vehicle it powers. This is especially crucial for electric vehicles, where concerns about driving range are prevalent. SoC estimation is frequently paired with state-of-health (SoH) estimation to assess the battery’s overall condition. This combination aids in forecasting the battery’s remaining lifespan and scheduling maintenance or replacement. In grid storage and renewable energy systems, precise SoC estimation aids in balancing energy supply and demand, ensuring dependable and efficient energy management. The above-mentioned discussion highlights the importance of the study which is carried out in this work. The existing works in the literature mainly use the extended Kalman filter (EKF) for the SoC estimation problem. It should be noted that the performance of the EKF degrades as the system’s non-linearity increases. Moreover, the existing battery management systems are complex and inherently involve high nonlinearities which further extend as these systems are expanded. For these reasons, the EKF may lose its applicability for applications demanding highly accurate SoC estimates. In this paper, therefore, we apply a more accurate cubature-quadrature Kalman filter (CQKF) to estimate the SoC of the Li-ion cell. The SoC estimates provided by the CQKF are more accurate than those provided by the EKF. In this regard, we first develop the completely observable equivalent circuit model (ECM) of a Li-ion cell by experimentally identifying the parameters of the cell model. The experimental study is carried out in a commercially available 2.5 Ah lithium iron phosphate (LFP) cell (A123 ANR26650M1-B). Subsequently, we apply the cubature quadrature Kalman filter (CQKF) for estimating the SoC of the considered Li-ion cell. We also perform a comparative analysis of the CQKF and the extended Kalman filter (EKF) based SoC estimation for the considered model. We further extended the analysis for the missing measurement case, where the actual measurement is intermittently lost or does not contain sufficient information for the state measurement. The efficacy of the estimation schemes is validated both experimentally and in simulation by computing several performance indexes. The simulation results show that, compared with the EKF, the implementation of the CQKF improves the SoC estimation accuracy significantly.