by Jožef Stefan Institute, Ljubljana, Slovenia (in Slovenian)
This paper investigates the use of machine learning methods for estimating the state of charge (SOC) of Li-ion batteries. Machine learning methods are evaluated in the real-world setting of an intelligent mobile home. The state of charge depends on battery capacity; the electrical current flowing in and out of the battery; and battery temperature. We first acquired historical measurements of current, voltage and temperature under regular operating conditions. We then applied the support vector machine method to estimate the state of battery charge and obtain accurate predictions for outperforming the existing voltage method.