Algorithm for Estimating SOC of Lithium-Ion Batteries Used for Electric Vehicles

Energy storage systems represents a critical role in power supply of electric vehicles (EV) and lithium-ion batteries is the most promising and dominant energy storage choice. Because of cell variations in a battery pack it is hard to obtain an accurate model to study the behavior of lithium-ion cells and in many situations different algorithms must be used to estimate state of charge. Several techniques for SOC estimation have been studied in the last years. These methods can be classified in three categories: direct methods, data-driven methods and state-observation methods. In this paper it is the intention to estimate state-of-charge (SOC) of lithium-ion battery using a state-observation method such as Kalman filter. The algorithm it is applied on an equivalent circuit model of a lithium-ion cell and after that can be extended to a pack of cells. Most of the time the performances are dependent on the measuring accuracy of current and open-loop calculation can lead to accumulated calculation errors due to uncertain disturbances from the practice application and lack of necessary corrective resolution. There are also other methods for estimating the SoC, like artificial neural networks, fuzzy logic, extended Kalman filtering based estimators but the main disadvantage of these methods is represented by the differences between each individual cell which are ignored most of the time. In this case for reliable and accurate results should take in consideration individual cell's state, make calculation and the lowest SoC should be the reference point.