A state-of-health estimation method of lithium-ion batteries based on multi-feature extracted from constant current charging curve

Abstract

State of Health (SOH) is critical for ensuring the safety and reliability of lithium-ion batteries. Incremental capacity analysis (ICA) method based on measurement data obtained during constant current (CC) charging is used for SOH estimation in the paper. Firstly, to improve the accuracy of SOH estimation in practical application, an improved feature extraction framework is proposed. It mainly includes three stages: data acquisition, data preprocessing and health indication generation. For depressing the noise, two methods are put forward. One is to reconstruct the charging voltage curve in the data preprocessing stage to avoid finding the wrong maximum of the IC curve. The other is to dispose of the outlier feature in the health indication generation stage. Secondly, a health indicator that can be used to characterize the fading of the batteries is proposed. It includes four features and they are the maximum value of the IC curve, the corresponding voltage, the energy and the capacity of a constant current (CC) charging interval determined by the maximum value of the IC curve. Finally, a support vector regression (SVR) model is built to connect the health indicator and the SOH of the battery. The experimental results show that the voltage curve reconstruction and the outlier feature disposition can weaken the influence aroused by the noise and the proposed health indicator can predict the SOH of the batteries with high precision.