Research on modeling and state of charge estimation for lithium-ion battery

Abstract

Based on the equivalent circuit model (ECM) of lithium-ion battery, this paper introduces an autoregressive and moving average with exogenous input (ARMAX) model. A recursive extended least square algorithm with forgetting factor is employed as the model parameter identification method, because there is a colored noise in operating process. Then HPPC test was conducted on a 20Ah LiFePO4 cell and the RC ECM parameters available were identified under different SOCs and different current rates. For higher accuracy SOC estimation and uncertainty reduction, strong tracking filter based on cubature Kalman framework (ST-CKF) is adopted as the SOC estimator to compensate for the drawback of nonlinear Kalman filter. The experimental results in UDDS test show that the ST-CKF estimator is more accurate than extended Kalman filter algorithm with the maximum estimated error about 1.8%.