Online Estimation of Battery Model Parameters and State of Charge Using Dual Time-Scaled Technique Without Open Circuit Voltage Experiment

The accuracy of estimating lithium battery internal parameters and state of charge (SOC) is closely related to the appropriate model and efficient algorithm. The real-time online estimation of the open-circuit voltage (OCV) model and the model parameters for multiple time scales are particularly significant. This article proposes an online dual time-scale recursive least squares (DTRLS) for closed-loop estimation of resistor–capacitor (
$RC$
) network parameters, OCV, and SOC of equivalent circuit model (ECM). The dynamic characteristics of the battery system are analyzed at multiple time scales. The model parameters are separated and decoupled to overcome the parameter deviation and numerical issue in the time domain of the classical recursive least squares (RLS) method. In the slow time scale, the OCV parameter is coestimated without additional offline open-loop tests to improve the accuracy of SOC. Under the Urban Dynamometer Driving Schedule (UDDS) test, the ECM parameters and terminal voltages estimated by the proposed DTRLS method are compared with that of classical forgetting factor RLS (FFRLS). The root-mean-squared error (RMSE) results show that the online closed-loop DTRLS method improves the estimation accuracy of terminal voltage, OCV, and SOC by 0.97%, 3.2%, and 2.06%, respectively.