A Comparative Study on Feature Extraction for Battery State of Health Estimation: Evaluating the Role of Open Circuit Voltage in LSTM Model

This paper investigates the addition of the open circuit voltage (OCV) parameter for state of health (SOH) estimation using a long short-term memory (LSTM) model. The battery degradation data of Oxford University is used in the investigation. The variable used for training the models are charging and discharging currents and voltage, times and temperatures. The comparison shows that the model incorporating OCV data performs better in estimation accuracy than the model without OCV.