Comparative study of curve determination methods for incremental capacity analysis and state of health estimation of lithium-ion battery. (June 2020)

• Record Type: Journal Article
• Authors: He, Jiangtao; Bian, Xiaolei; Liu, Longcheng; Wei, Zhongbao; Yan, Fengjun
• Is Part Of: Journal of energy storage. Volume 29(2020)
• Publication Date: 2020-06
• Subjects: Lithium ion battery -- State of health -- Incremental capacity analysis -- Gaussian -- Lorentzian; Energy storage -- Periodicals; Energy storage -- Research -- Periodicals; 621.3126
• DOI: 10.1016/j.est.2020.101400
• Languages: English
• ISSNs: 2352-152X

Abstract

Highlights: 3 existing voltage-capacity (VC) models and 3 newly proposed modified models are thoroughly compared for incremental capacity analysis (ICA) and state of health (SOH) estimation. The feature of interests (FOIs) extracted from VC models are analyzed with respect to their linearity with the capacity, offering deep insights into more straightforward SOH estimation for LIB. VC models are compared with widely-used pre-filtering methods in terms of the curve deformation and aging consistency. Abstract: Incremental capacity analysis (ICA) is a favorable candidate for state of health (SOH) estimation of lithium-ion battery (LIB). Although abundant works have been carried out on the ICA-based methods, a comprehensive comparison of them to clarify the application boundary is still lacking. Moreover, more efficient method for extracting more informative features of interest (FOIs) for SOH estimation is less explored. Motivated by this, this paper performs a comparative study over the filtering-based and the voltage-capacity (VC) model-based ICA methods with respect to the IC fitting accuracy, robustness to aging and the computing cost. In this framework, a set of novel FOIs different from traditional ones are captured along with the parameterization of VC models. Comparative results reveal the optimality of revised Lorentzian VC model with three peaks (RL-VC-3) for both LiFePO4 (LFP) and LiNi1/3 Co1/3 Mn1/3 O2 (NCM) battery. The mean relative errors of capacity modeling are 0.34% and 0.15%, respectively. The newly captured FOIs have been further validated with high linearities with the reference capacity, offering deep insights into more straightforward SOH estimation for LIB. Illustrative case studies suggest that particular FOIs can offer accurate SOH estimation with absolute error of 0.079% and 0.661% respectively for the LFP and NCM battery.