State of charge estimation of lithium-ion batteries based on second-order adaptive extended Kalman filter with correspondence analysis

Accurate estimation of the state of charge (SOC) of a lithium-ion battery is important to ensure the safe operation of the battery management system. Adaptive extended Kalman filter (AEKF) is used to estimate SOC, which approximates nonlinear function to linear function by first-order Taylor expansion with large truncation error. Therefore, the second-order AEKF is proposed to reduce the truncation error and improve the accuracy of SOC estimation. Since the estimation accuracy of second-order AEKF is also affected by the sliding window length (SWL), correspondence analysis is proposed in this paper to verify the correlation between SWL and algorithm errors and obtain a reasonable SWL parameter value, which helps to ensure that the algorithm has higher accuracy in SOC estimation under the condition that the value of SWL is not changed when the working condition changes. To substantiate the efficacy of the algorithm outlined in this paper, data sets collected from different sources and at various temperatures are employed. The experimental results obtained through meticulous analysis demonstrate that the second-order AEKF proposed in this study excels in terms of estimation accuracy and robustness.