An appropriate battery model is crucial for accurate state of charge (SOC) estimation of lithium-ion batteries. A complex battery model can improve the accuracy, however, it leads to high computational cost. In this paper, the second-order resistor-capacitor (RC) model is selected due to its superiority in good balance between accuracy and complexity. Different factors, including RC parameters of battery model, relationship between open circuit voltage (OCV) and SOC, and measurement noise are investigated to extract the ones that significantly affect the accuracy of SOC estimation based on Kalman filters. The results revealed that in comparison with the resistors, the polarization capacitors have no obvious influence on the SOC estimation. In addition, the voltage measurement noise has greater influence on the SOC estimation compared with the current measurement noise. These results can be potentially used to guide us in deciding which parameters should be separately identified online and offline so as to achieve a proper balance between model accuracy and complexity.
Keywords lithium-ion battery, state of charge, model parameters, Kalman filters