State of Charge Estimation for Lithium-Ion Batteries Using a Temperature-Based Equivalent Circuit Model

This study investigates battery state-of-charge (SOC) estimation under different temperature conditions. A battery modeling approach is developed aiming to improve the accuracy of the SOC estimation when ambient temperature is taken into account. Firstly, a widely used equivalent circuit model with the one-order resistance-capacitor (RC) network is modified to capture battery dynamics at different temperatures. Secondly, since the open-circuit voltage verse SOC (OCV-SOC) incorporated into the battery model is also influenced by the temperature, OCV-SOC-Temperature (OCV-SOC-T) table is constructed to replace the original table based on our experimental data. The experiments with two dynamic load tests, dynamic stress test (DST) and federal urban driving schedule (FUDS) are run on the battery. The purpose of DST profile is to identify the battery model, while FUDS data is used to emulate the operation conditions and evaluate the performance of our proposed model by unscented Kalman filtering. Finally, the comparative results indicate that our temperature-based model provide more accurate SOC estimation with root mean square estimated errors than the original model without regard to temperature dependence.