A sparse least squares support vector machine used for SOC estimation of Li-ion Batteries

Abstract

Abstract Li-ion batteries have been widely used in electric vehicles, power systems and home electronics products. Accurate real-time state-of-charge (SOC) estimation is a key function in the battery management systems to improve the operation safety, prolong the life span and increase the performance of Li-ion batteries. Kalman Filter has shown to be a very efficient method to estimate the battery SOC. However, the battery models are often built off-line in the literature. In this paper, a least squares support vector machine (LS-SVM) model trained with a small set of samples is applied to capture the dynamic characteristics of Li-ion batteries , enabling the online application of the modelling approach. In order to improve the model performance of battery model, a sparse LS-SVM model is first built by a fast recursive algorithm. Then, the batteries SOC is estimated using an unscented Kalman filter (UKF) based on the sparse LS-SVM battery dynamic model. Simulation results on the Hybrid Pulse Power Characteristic (HPPC) test data and the Federal Urban Drive Schedule (FUDS) test data confirm that the proposed approach can produce simplified yet more accurate model.