Accuracy improvement of SOC estimation in lithium-ion batteries

Abstract

Scheduling Lithium-Ion batteries for energy storage applications in power systems requires an accurate estimate of their state of charge (SOC). The Coulomb counting method is popular in the industry but remains inaccurate.This paper presents an intelligent technique for the SOC estimation in Lithium-Ion batteries. The model is developed offline using adaptive neuro-fuzzy inference systems (ANFIS). It considers the cell nonlinear characteristics, as supplied by the manufacturer, which provide the relationship between the cell SOC and open-circuit voltage (OCV) at different temperatures. The manufacturer data are used to model the cell characteristics by ANFIS in order to yield the cell SOC at any arbitrary OCV and temperature within the given range. The pack SOC is accordingly estimated.For the purposes of comparison, the Coulomb counting method is used at the cell level, rather than the pack level, to estimate the SOC of the battery. Laboratory experiments are conducted on a 5.3kWh battery module where measured SOC is compared to Coulomb counting computations at the cell and pack levels. Results show distinct superiority for the proposed ANFIS technique over the traditional Coulomb counting method.