Comparison of an Equivalent Circuit and Data Driven Battery Model for EV Applications

Both electric and hybrid vehicles (HEVs and EVs) employ high-voltage battery packs, so a robust and dependable battery management system (BMS) is necessary to ensure a vehicle's safe and dependable operation. For proper functioning of the BMS, battery state estimation is a key challenge. This paper presents an equivalent circuit and a data driven model, evaluating and contrasting their appropriateness in terms of accuracy as compared to an experimental battery. Furthermore, the state of charge (SOC) of battery is estimated, first with the help of the 2RC equivalent circuit model (ECM), then with the help of an artificial neural network (ANN). A pulsed discharge test is carried out and the SOC and voltage values are noted. The parameters of the 2RC model are tuned for improving the accuracy. The model with different parameters is compared with an experimental battery. With the data driven model, a better accuracy is achieved in comparison with experimental data. With the ANN the mean square error (MSE) of 0.8715% is achieved, whereas with the ECM an MSE of 2.87% is achieved.