State of Charge Estimation of Lithium-Ion Batteries Using an Extended Kalman Filter with Optimized OCV Modeling

The Battery Management System (BMS) performs tasks in order to ensure the safe and reliable operation of lithiumion battery packs in electric vehicles (EVs). To ensure this, a primary function of the BMS is to perform state estimation for the state of charge (SOC), among other states. SOC defines how much capacity is left in a battery. Due to the complex nature of electrochemical batteries, the SOC must be estimated with respect to the current, voltage, and temperature of the cell, which requires advanced estimation techniques. A commonly used method is a form of the Kalman filter as the filter is able to account for all of these factors and predict a SOC value with accuracy compatible with automotive specifications. This approach relies on a battery equivalent circuit models (ECMs) in order to perform the estimation. In this paper, the performance of Kalman filter-based SOC estimation algorithms will be assessed considering different-order ECMs and different model parameter approximations. This will be done by performing a statistical analysis on the absolute error of the terminal voltage approximation achieved with Kalman filter algorithms.