基于数据驱动方法在液态金属电池容量骤降前的提前预警

The liquid metal battery (LMB) has gained attention as a novel energy storage technology due to its exceptional safety and long lifespan. Analyzing the aging trajectory, particularly capacity plunge process, is crucial for understanding its aging mechanism and enabling effective health diagnosis. However, it often lacks clear warning signs prior to the capacity plunge, hindering accurate capacity prediction and proactive measures in real-world applications. To address this challenge, this study proposes a data-driven method that effectively quantifies the capacity various in LMBs, enabling advance warning before the occurrence of a capacity plunge. Firstly, the Empirical Mode Decomposition method decomposes the capacity data into several components, representing different characteristics of LMBs. Subsequently, a Gaussian mixture model is applied to construct the feature distribution for each component, which exhibits a hump-shaped pattern. Finally, The Jensen-Shannon divergence is used to quantify the divergence of these humps, offering a concrete measure of LMB health and aging progression. The proposed method integrating both aging analysis and data-driven insights provides an advance warning of the capacity plunge, offering a reliable foundation for future lifespan prediction and fault diagnosis of LMBs.