State of health estimation of lithium-ion battery by removing model redundancy through aging mechanism

Abstract

Accurate estimation of lithium-ion battery state of health (SOH) is imperative to maintain the safe operation of electric vehicles. Multiple health indicators (HIs) from constant-current-constant-voltage (CC-CV) charging cycle and nonlinear models are used to estimate the SOH of lithium-ion batteries in the recent data-driven methods. Due to not considering the physical relationship between HIs and SOH, redundancy of HIs and models are often existed. In order to reflect the physical characteristics of the battery and remove the redundancy of the model, a strong linear relationship between constant-current charging time and SOH is testified by analyzing the aging mechanism of lithium-ion batteries and a linear regression model is used to estimate the SOH of lithium-ion batteries using constant-current charging time as the only health indicator. The experimental results show that the average error of 8 batteries is within 1.2%, and the average training and testing time is within 0.03 s. This work aims to focus on the mechanistic analysis of the battery to select HIs appropriately and to improve the SOH estimation efficiently.