UC Irvine

The state of charge (SOC) estimation of lithium-ion battery is a core function of the battery management system, which reflects the remaining charge of the battery and is an important parameter for predicting the remaining driving range of an electric vehicle. The accuracy of the battery SOC estimation depends on the precision of the battery model and the validity of the estimation method. In practice, the complex electrochemical characteristics inside the battery and the dynamic changes of the external environment make it difficult to estimate the battery SOC with high accuracy. To enable high accuracy SOC estimation, this paper proposes a method to estimate battery SOC based on the combination of simplified electrochemical model and Bi-directional gated recurrent unit (Bi-GRU) network. In the model simulation, we compare the simplified electrochemical model, the pseudo-two-dimensional model, and the equivalent circuit model and verify that the simplified electrochemical model is more precise than the equivalent circuit model and less precise than the pseudo-two-dimensional model. In the SOC estimation, we compare our proposed method, ampere hour integral, particle surface utilization rate based on simplified electrochemical model, and extended Kalman filter based on equivalent circuit model. The results show that the accuracy of SOC estimation based on the proposed method is as high as 99.97% and this method outperforms the other methods studied in this work. This work reveals the important role of data-driven and simplified models in the field of battery cloud digital twin and provides a promising strategy for real-time state estimation for next-generation cell-level battery management systems.