Lithium Battery State-of-Health Estimation via Differential Thermal Voltammetry With Gaussian Process Regression

Accurate state-of-health estimation can give valuable guidelines for improving the reliability and safety of energy storage system. In this article, a novel battery degradation tracking method is proposed through the fusion of significant health features with Gaussian process regression (GPR). First, an advanced filter method is used to smooth differential thermal voltammetry (DTV) curves. Thereafter, considering the relationship between battery degradation and DTV curves, some health factors are extracted from DTV curves. In this article, these health factors involve different dimensions of the DTV curve, including peak position, peak, and valley values. Third, a correlation analysis method is employed to select four high-quality features from health factors, which are fed into GPR to learn and establish a battery degradation model. Finally, the estimation accuracy, robustness, and reliability of the proposed model are verified using four batteries with different aging test conditions and health levels. The results demonstrate that the proposed model can provide accurate battery health status forecasting.

Automated summary

Accurate state-of-health estimation is crucial for enhancing the reliability and safety of energy storage systems. This article proposes a novel battery degradation tracking method that incorporates significant health features with Gaussian process regression. By utilizing advanced filter methods to smooth differential thermal voltammetry curves, extracting health factors, and selecting high-quality features through correlation analysis, a reliable battery degradation model is established and verified. The results show that the proposed model can effectively forecast battery health status.