SOC Estimation for Lithium-Ion Batteries Based on the AEKF-SVM Algorithm

Authors:
  • LZLei Zhang
    Lei Zhang
  • QSQianqian Song
    Qianqian Song
  • JLJie Li
    Jie Li

Abstract
In light of societal progress, energy has emerged as a critical concern, and new energy sources are poised to become the energy mainstream. Among these developments, lithium-ion batteries play a pivotal role in the new energy industry. Accurately estimating the State of Charge (SOC) of lithium-ion batteries is essential for the proper functioning of devices like electric vehicles. However, the SOC of lithium-ion batteries cannot be measured directly by instruments and can only be estimated by measurable variables. This paper proposes a new algorithm for accurate SOC estimation using a combination of adaptive extended Kalman filtering (AEKF) and support vector machine (SVM) algorithms, taking into account the characteristics of lithium-ion batteries. The AEKF algorithm was employed to estimate the State of Charge (SOC) under the Beijing Bus Dynamic Stress Test (BBDST) condition. By leveraging the adaptive noise covariance advantage of the AEKF algorithm, a model that is more fitting for the lithium-ion battery was obtained. Subsequently, the SOC for the Hybrid Pulse Power Characterization (HPPC) and Dynamic Stress Test (DST) conditions were predicted. Experimental results revealed that the SVM-AEKF algorithm, when used for SOC estimation, resulted in a maximum error of 0.037% under the HPPC condition, marking an improvement of 8.9% compared to the AEKF algorithm. Under the DST condition, the maximum error was 0.335%, indicating a 6.9% improvement over the AEKF algorithm. These results underscore the potential of the SVM-AEKF algorithm in accurately estimating SOC, thereby holding promise for practical applications.