A Feature-Based Data-Driven Approach for State-of-Health Estimation of Satellite Lithium-ion Batteries

The present work introduces a data-driven framework for evaluating the SOH of lithium-ion batteries used in satellites, addressing the unique challenges posed by the space environment. A robust methodology is proposed for indirect SOH estimation based on optimal health indicator (HI) extraction. By leveraging correlation analysis, feature redundancy is minimized and its relevance to capacity prediction is maximized. This approach facilitates the generation of an optimal set of features that ultimately enhances the accuracy of the SOH prediction framework. The feasibility and validity of the proposed methods are demonstrated using a lithium-ion battery test dataset from the NASA prognostic center.