Battery modeling and Kalman filter-based State-of-Charge estimation for a race car application

Abstract

This paper investigates a method for building a lithium-ion polymer battery model, which can be used as part of a model-based estimation technique to estimate battery State-of-Charge for a Formula Student electric race car application. The modeling strategy is based on developing an equivalent circuit, which can capture the behavior of the battery dynamics experienced at the competitions. The equivalent circuit model accounts for steady-state and dynamic contributions due to open-circuit voltage, internal resistance, and polarization dynamics. Using experimental cell tests that are representative of the current load experienced by the batteries, and Simulink Parameter Estimation to parameterize the equivalent circuit, a root-mean-square modeling error of 8.0 mV was obtained. Utilizing the model as part of an Extended Kalman Filter to estimate State-of-Charge, a root-mean-square estimation error of 0.58%, and a maximum absolute estimation error of 2.37% were achieved.