A syncretic state-of-charge estimator for LiFePO4 batteries leveraging expansion force. (June 2022)

• Record Type: Journal Article
• Authors: Xu, Peipei; Li, Junqiu; Xue, Qiao; Sun, Fengchun
• Is Part Of: Journal of energy storage. Volume 50(2022)
• Publication Date: 2022-06
• Subjects: LiFePO4 batteries -- SoC estimation -- Expansion force -- Adaptive unscented Kalman filters (AUKF) -- Least-square support vector machines (LSSVM); Energy storage -- Periodicals; Energy storage -- Research -- Periodicals; 621.3126
• DOI: 10.1016/j.est.2022.104559
• Languages: English
• ISSNs: 2352-152X

Abstract

Abstract: State of Charge (SoC) estimation for LiFePO4 (LFP) batteries is particularly challenging due to the flat open circuit voltage (OCV) characteristics. This paper proposes a novel method of SoC estimation for LFP batteries based on expansion force which has not been investigated in the existing literature. After a series of experiments, it is found that the expansion force is more sensitive to SoC than voltage and independent of the dynamic current. However, the estimation work is still technically challenging because of the non-monotonic relationship between expansion force and SoC. To cope with it, an expansion force model based on the least-square support vector machine (LSSVM) method is firstly exploited, which is employed as the measurement equation of adaptive unscented Kalman filters (AUKF). Meanwhile, the moving window method is applied to enhance the adaptability of the established model and then an accurate SoC estimation result is attained. Finally, the proposed method is evaluated via sufficient experiment data covering different temperatures and constraint conditions. Experimental results show that the root means square errors of measure equation and SoC estimation can be bounded within 1% and 0.54%. In conclusion, the proposed method provides a new perspective for SoC estimation of LFP batteries. Highlights: The evolution of battery expansion force under different dynamic current profiles is investigated. A data-driven method is proposed to predict the expansion force based on the LSSVM model. The moving window method is applied to enhance the adaptability of the established model. The precise SOC estimation is attained for LiFePO4 batteries based on the predicted expansion force and the AUKF algorithm. The estimation accuracy is evaluated under different and temperatures and constraint conditions.