State of health estimation for Li-ion battery via partial incremental capacity analysis based on support vector regression

Precise battery capacity estimation and monitoring are of extreme importance for the future intelligent battery management system. The primary technical issues result from the absence of enough cognition for battery aging mechanism and effective modeling in complex application scenarios. Synthesis theoretical analysis and engineering application, incremental capacity analysis approach may be accessible in actual operation. This paper proposes a data-driven prediction technique, support vector regression for establishing a battery degradation model, which estimates battery capacity by partial incremental capacity curves. Firstly, the advanced filter algorithms are utilized to smooth incremental capacity curves and then a peak fitting technique is applied to decompose the smooth curves. The battery health features are extracted from decomposed incremental capacity curves as training datasets. Using different sizes of training datasets, three battery degradation models are established based on the support vectors regression algorithm. The performances of the proposed models are comparison analyses for each testing dataset. The aging datasets are collected from other three batteries applied to extensively verify the proposed method. Quantitatively, mean absolute errors (MAEs) and root mean square errors (RMSEs) of the three models are both limited to 2%. Otherwise, the accuracy of Model3 is improved about 30% in MAEs and RMSEs. • Two filter methods are proposed to smooth the incremental capacity curves. • Novel peak fitting based method decomposes the smooth incremental capacity curves. • Health factors extract from areas, peaks and heights of the decomposed curves. • SVR-based degradation models learn from different sizes of training datasets. • Two type batteries verify and evaluate the performances of the proposed method.