State of health estimation based on modified Gaussian process regression for lithium-ion batteries

State of health (SOH) estimation of lithium-ion batteries is a challenging and crucial task for consumer electronics, electric vehicles, and micro-rids. This study presents a data-driven battery SOH estimation method based on a novel integrated Gaussian process regression (GPR) model. First, the aging characteristics of batteries are analyzed from multiple perspectives, and three health indicators (HIs) are extracted from battery charging and discharging curves. Then, the Pearson correlation analysis method is used to quantitatively analyze the correlation between the selected HIs and SOH. Next, a novel compound kernel function is proposed for battery SOH estimation, and different pairs of mean function and kernel function chosen from four mean functions and sixteen kernel functions are used to construct GPR models, and their estimation accuracy is compared subsequently. Finally, four different batteries with various initial health conditions from the NASA battery dataset are used to verify the performance of the proposed method. Experiments show that the method proposed in this paper has satisfactory estimation results in terms of accuracy, generalization ability, and robustness. Specifically, its estimated mean-absolute-error (MAE) and root-mean-square-error (RMSE) is only 1.7%, and 2.41%, respectively.

Automated summary

This study proposes a data-driven method for state of health estimation of lithium-ion batteries. The method utilizes Gaussian process regression models based on the analysis of aging characteristics and extraction of health indicators from battery charging and discharging curves. The experiments demonstrate that the proposed method achieves satisfactory estimation results in terms of accuracy, generalization ability, and robustness.