Online Estimation of Lithium Battery SOH Based on Incremental LS-SVR Algorithm

Abstract

The accurate estimation of state of health (SOH) of lithium-ion batteries is the key of battery health management. This paper proposes an incremental LS-SVR algorithm with the ability to update the model online. The health factor is derived from the battery charging data with constant current and constant voltage mode by incremental capacity analysis. An initial LS-SVR model is trained on the offline data set, and then applied online on other data sets with the same experimental conditions. In the process of online application, the model will be updated online every certain cycle. The experimental results show that the incremental LS-SVR algorithm adopted in this paper has higher accuracy than the direct application of offline model, and it can well solve the problem of non-independent and identically distributed samples in different datasets, which bring certain practical value.