State-of-Charge Estimation of the Battery Based on Improved Equivalent Circuit Model and Adaptive Extended Kalman Filter

The fuel cell propulsion system is one of the potential development directions for future green aviation. Due to the slow dynamic response of fuel cells, they need to be used together with lithium-ion batteries to form a hybrid power system. The implementation of most functions of lithium-ion batteries requires the State of Charge (SOC) as the basis, and accurate estimation of SOC is of great significance for the stable operation of the battery. Taking into account the requirements for modeling accuracy and complexity, this paper presents an improved second-order RC Equivalent Circuit Model (ECM) that incorporates hysteresis effects. The model parameters were identified using data obtained from Hybrid Pulse Power Characteristic (HPPC) tests. Recognizing the limitations of the Extended Kalman Filter (EKF) algorithm in terms of its interference rejection capabilities, this paper introduces a noise correction matrix via the windowing method to develop an Adaptive Extended Kalman Filter (AEKF). A simulation model was constructed to validate the SOC estimation performance of the proposed algorithm under various operating conditions. The simulation results demonstrate that the AEKF exhibits superior convergence and interference rejection capabilities, with SOC estimation error of less than 1%, meeting the required precision standards.