Prediction of State of Charge of Battery in Electric Vehicles Using Artificial Neural Network

The trajectory of today’s world is towards emission-free energy sources. The result is a huge growth in the number of Electric Vehicles (EVs) experienced around the world. However, the batteries in EVs must be properly operated to extend their lifetime for which an accurate estimation of the State of Charge (SOC) of the cell is crucial. The conventional methods of SOC estimation are either less accurate or computationally complex. This paper presents a method of estimating the SOC of a Li-ion cell used in EVs employing an Artificial Neural Network (ANN) model which is tested in a 1-RC Equivalent Circuit Model (ECM) of a cell in MATLAB, SIMULINK. The model uses the measured current, voltage, and temperature, and predicts SOC based on these quantities. The first input feature for the ANN model is the power, which is calculated by multiplying the current and voltage. The average voltage and current over 600 timestep intervals are used as the second and third features, respectively. Additionally, the fourth input feature is the temperature. The performance of the model is tested against experimental data.