A Hierarchical Approach for Finite-Time H-$\infty$ State-of-Charge Observer and Probabilistic Lifetime Prediction of Lithium-Ion Batteries

Abstract

Accurate state-of-charge (SOC) estimation and lifetime prognosis of lithium-ion batteries are of great significance for reliable operations of energy storage systems. This paper proposes a novel two-layer hierarchical approach for online SOC estimation and remaining-useful-life (RUL) prediction based on a robust observer and Gaussian-process-regression (GPR). At the bottom layer, an equivalent-circuit model is first developed to describe battery dynamics. Second, a combination method of a recursive least square method and a finite time H- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\infty$</tex-math></inline-formula> observer is designed to estimate battery open-circuit-voltage (OCV) and SOC through stability and robustness analysis. Next, the estimated OCV and SOC are fed into the top layer to generate the incremental-capacity-analysis-based aging feature, through which a robust signature associated with battery aging is identified. The feature is further employed for RUL prediction based on GPR. The salient advantages of the proposed approach are that it can provide robust parameter estimation in a given finite-time interval, and the GPR-based RUL prediction can tackle long-term uncertainties in a principled Bayesian manner. Theoretical analysis and experimental results demonstrate the effectiveness of the proposed SOC observer and lifetime prediction methods.