Influence of Initialisation Parameter in Extended Kalman Filter on State of Charge Estimation

Abstract:
Batteries are widely used for varied applications, and yet more applications are being discovered. Accurate parameter and state estimation of batteries remain challenging due to the electrochemical nature and varied degradation phenomenon. Hence, novel estimation algorithms are being researched worldwide to increase accuracy and interpretability of these estimations. Irrespective of many new algorithms, Extended Kalman Filters remain an optimal choice for performing parameter and state estimations. One of the most important values that impact the accuracy is the choice of covariance. The literature reveals the relevance, such as limitations on selecting a high or low value but does not define any initialisation methods. Further, there is a lack of studies showing the impact of the covariance initialisation on the results. This manuscript will underline the relevance of covariance in their impact on the accuracy of the expected results. The results will be presented based on the laboratory studies performed on different types of cells of varied electrochemistry and rating.