A novel approach to state of charge estimation using extended Kalman filtering for lithium-ion batteries in electric vehicles

This paper proposed a novel approach to state-of-charge (SoC) estimation of the lithium-ion batteries (LiBs) used in electric vehicles (EVs) based on the extended Kalman filtering (EKF). An improved lumped parameter model was developed for describing the dynamic behavior of the LiBs with an optimized open circuit voltage. This improved approach can reduces model error effectively. Other model parameters were identified via the genetic algorithm (GA) to optimizes the polarization time constant. Experimental and simulation results with two kinds of dynamic cycles show that, compared to the commonly used coulomb counting method, the EFK based SoC estimation algorithm is more precise. The proposed methodology can resolve the deficiency of coulomb counting method. The coulomb counting method fails to correct the erroneous initial SoC and is prone to cause greater accumulated error. In contrast, the proposed novel SoC estimation approach can accurately project the SoC trajectory. It employs real-time measurements of battery current and voltage. This approach then can be applied conveniently to battery management system in commercial electric vehicles.