State-of-charge estimation of lithium-ion batteries from a gas-liquid dynamics model including the direct temperature input

Abstract

Accurate estimation of the online state-of-charge of the battery is one of the crucial issues in a battery management system. An improved gas-liquid dynamics model is proposed to simulate the physicochemical behaviors of a lithium-ion battery, such as the electron transference, terminal voltage lagging open-circuit voltage, Li+ diffusion and equilibrium, and ohmic effect. The state-of-charge estimation equations with open-circuit voltage and temperature iterations are deduced according to this model. A robust method of online state-of-charge estimation is elaborated to achieve the goal of quickly eliminating the incorrect initial state-of-charge and temperature. The soft-packing, cylindrical and prismatic Li(NiMnCo)O2 batteries are used for experimental verification under the constant current test, the Dynamic Stress Test cycle and the Urban Dynamometer Driving Schedule cycle, respectively. Furthermore, the study of the effect of the sampling period on estimation accuracy and the sensitivity analysis of the parameter is carried out. For the three types of batteries, the results show that the maximum SOC estimation error of the proposed method is 2.78%. The proposed method has strong robustness against the initial errors of 100% state-of-charge and ±20 K temperature.