State-of-Charge Estimation of Lithium Battery Using Adaptive Extended Kalman Filter With LMAT Driven Parameter Estimation

Accurate estimation of State-of-charge (SOC) act as a decisive factor for the management of lithium batteries, especially in applications like energy storage system (ESS), electric vehicles (EV) and portable electronics devices. Recently, model-based SOC estimation gained popularity due to their easy implantation and capability to capture the dynamic characteristics of the lithium battery. However, in model based SOC estimation, the battery model parameter tends to vary with the change in operating temperature. This variation in battery model parameter, affects the estimation accuracy of the filter significantly. In view of this, a least mean absolute third (LMAT) based adaptive extended Kalman filter (AEKF) is designed and proposed in this article. In the proposed method, the dynamic model parameter of the battery are estimated with a computationally efficient LMAT-based estimator and the battery SOC is estimated by adaptive correction of noise covariance matrix using AEKF. The SOC estimation accuracy of the proposed LMAT-based AEKF is examined by applying load current of the LA92 and US06 driving schedule at 0°C and 25°C respectively. Further, the performance indices of the proposed estimator are compared with other Kalman filter based SOC estimators available in the literature. The results confirm high SOC estimation accuracy of the proposed LMAT based AEKF compared to other KF based estimators with a root mean square error of less than 2% under dynamic operating condition.