Adaptive Extended Kalman Filtering Based State-of-Charge and Voltage Estimation of a Lithium-Ion Battery Using a Fractional-Order Model

The most crucial parameter for the effective operation of a battery management system (BMS) is the battery terminal voltage, which is usually used for monitoring the threshold faults, state-of-charge (SOC) estimation, and cell balancing. Based on the literature, the battery fractional-order models (FOMs) provide much better accuracy compared with the integer-order models (IOMs). For battery SOC and voltage estimation, the extended Kalman filter (EKF) is a popular algorithm. However, it has certain limitations due to its nonadaptive nature of the noise covariances, which may result in erroneous estimates. Therefore, an adaptive EKF (AEKF) algorithm can be a better choice which overcomes the limitations of an EKF. In this article, both the FOM and AEKF are combined together in a fractional-order AEKF (FO-AEKF), where the adaptive nature of the process and measurement noise covariances in the AEKF attempts to improve the estimation accuracy. A comparison between IOM and various FOMs has been shown and validated on two standard datasets, i.e., constant current discharge schedule (CCDS) from NASA and urban dynamometer driving schedule (UDDS). Based on the comparative analysis, it has been observed that the FO-AEKF corresponding to the FOM with Warburg impedance (FOMW) provides the best accuracy among all the methods considered, in terms of the root mean squared error (RMSE).