State of Charge Estimation for Lithium-ion Batteries using Extreme Learning Machine and Extended Kalman Filter

Abstract

State of charge (SoC) estimation is one of the most important functions for battery management systems (BMSs). Due to the complex electrochemical characteristics of Lithium-ion batteries (LIBs), accurate SoC estimation remain challenges. To take full advantage of the widely used model-based methods and data-driven methods, an extreme learning machine-extended Kalman filter (ELM-EKF)-based method is proposed for SoC estimation in this paper. The ELM is utilized to establish an accurate LIBs model first. Then, the trained ELM model is combined with the EKF algorithm for SoC estimation. The proposed ELM-EKF-based SoC estimation method is validated and compared with the traditional equivalent circuit model-EKF (ECM-EKF)-based method under Federal Urban Driving Schedule (FUDS) driving cycles at three different temperatures. The results prove that the ELM model have better voltage-tracking capability than the ECM model while the ELM-EKF-based SoC estimation algorithm can achieve higher estimation accuracy than the ECM-EKF-based method.