Comparison of Neural Network and System Identification approaches for battery modeling and State of Charge estimation

A comparison of battery modeling results and state-of-charge estimation using neural networks and system identification approaches was conducted. These methods were applied to both simulation results and real battery measurements. Simulation data was used to determine the structure of the neural network and a linear regression model. Real battery data, consisting of two sets from two different batteries, was used to train and validate the models. The outcome of this study is black-box models derived from these two different approaches.