A novel method for state of charge estimation of lithium-ion batteries at low-temperatures

The low temperature environment poses a significant challenge to the application of electric vehicles (EVs). At low temperatures, the dynamic characteristics inside the battery become significantly different from those in the temperature range of 10–40 °C, resulting in high uncertainties in the estimation of state of charge (SOC). Experimental studies on two types of lithium-ion batteries have found that due to changes in battery polarization characteristics at low temperatures, the open circuit voltage (OCV) identified by the commonly used equivalent circuit models and parameter identification methods becomes more distorted. This is the reason for the failure of most SOC estimation methods based on OCV-SOC mapping. A part of polarization voltage is incorrectly involved in the OCV by online parameter identification. Based on this phenomenon, a novel method is proposed to achieve accurate SOC estimation at low temperatures by compensating this part of polarization voltage. The compensation voltage is calculated by a function, which is identified from experimental data using genetic algorithm. The validation against experimental results demonstrates that the proposed method can achieve a root mean square error and mean absolute error of less than 3 % for the SOC estimation in temperatures down to −20 °C. Moreover, this method only needs experimental data of dynamic operating conditions measured at two temperatures which cover most of the battery's working temperature range. And its computational complexity is low, making it suitable for onboard applications.