State of health estimation of lithium-ion batteries based on data-driven methods with a selected charging voltage interval

Abstract

Accurate state of health (SOH) estimation of lithium-ion batteries is of great importance for achieving efficient energy management in the overall battery energy storage system. Traditional data-driven methods for the SOH estimation of lithium-ion batteries usually require an enormous amount of data from the whole charging phase, which leads to poor performance in both computational efficiency and computational cost. To address this issue, this paper proposes the SOH estimation methods for lithium-ion batteries based on the limited data from a selected charging voltage interval. First, this study uses incremental capacity curves and Pearson correlation analysis to select an optimal and limited charging voltage interval that is the most relevant to lithium-ion battery degradation. Then, the SOH estimation based on two typical data-driven methods, including random forest regression (RFR) and support vector regression (SVR), would be implemented with the selected charging voltage interval. Results show that both the RFR and the SVR methods can achieve excellent accuracy, while each has its own irreplaceable advantages. However, compared with other voltage intervals using the two data-driven methods, the corresponding SOH estimation with the selected charging voltage interval shows the best performance. Hence, the data-driven methods based on the selected charging voltage interval have significant potential and advantages in the field of lithium-ion battery SOH estimation.

Keywords

• lithium-ion battery
• state of health
• data-driven methods
• random forest regression
• support vector regression