Extended Kalman Filter for state-of-charge estimation in electric vehicles battery packs

Abstract

Efficacious work of the Li-ion battery (LIB) is strongly influenced by several factors as temperature, nominal voltage, capacity and charge/discharge current rate, determining its health and state-of-charge (SOC). SOC is an indicator mirroring the available charge stored in the battery relative to its maximum capacity when new, resembling the fuel gauge in the conventional, internal combustion engine (ICE) vehicles. As this parameter defines rather a condition than a physical quantity, direct measurement using classical means is not feasible. Hence, one needs to “precisely guess” the actual available charge within the battery, in order to operate the battery pack effectively. Among the many methods employed for estimating the batteries' SOC, the Extended Kalman Filter (EKF) stands out due to its intrinsic predictor-corrector mechanism, which makes the filter insensitive to various modeling deficiencies like poor impromptu initialization or noise. This paper presents a step-by-step design procedure of such a filter, based on a 2ndorder Thévenin battery equivalent circuit model (ECM). The high performance of the filter estimator is validated against laboratory measurements.