Nonlinear modeling and SOC estimation of lithium-ion batteries based on block-oriented structures

The accuracy of lithium-ion battery model directly affects the safety and reliability of the battery system. The traditional equivalent circuit model (ECM) cannot reflect the nonlinear characteristics inside the batteries. This paper proposes a nonlinear ECM (NL-ECM) based on block-oriented structures to accurately describe and quantify the characteristics of lithium-ion battery. A few structured configurations of the NL-ECM are established by combining dynamic linear and static nonlinear modules. The NL-ECM based on the Hammerstein-CARMA configuration is taken as the research object, and the multi-strategy improved coyote optimization algorithm is used to identify its parameters. Experimental tests and simulation verification are carried out at different temperatures. The results show that the model accuracy of the NL-ECM is more than 70% higher than that of the ECM. Based on this model, the state space and measurement equations are established, and the adaptive extended Kalman filter algorithm is introduced for state of charge (SOC) estimation. The RMSE and MAE of the SOC estimation results are around 0.6% with no initial error in SOC, which are reduced by more than 40% compared with those of the ECM. At the same time, the AEKF algorithm based on the NL-ECM still exhibits good robustness even in the presence of initial errors in SOC.