Health EquiMetrics for Battery Assessment– A Novel Metric for End-of-Life Evaluation Beyond State-of-Health

Accurately assessing lithium-ion battery (LiB) health is essential for optimizing battery-powered system performance and enabling second-life applications. Currently, battery health prognosis primarily relies on the state-of-health (SOH) metric, which is closely linked to individual capacity metrics and discharge characteristics. However, this reliance limits comparative analysis and can lead to incomplete or misleading conclusions about battery health and lifespan, potentially affecting their suitability for various applications. To overcome this limitation, we introduce Health EquiMetrics (HEM), a metric that benchmarks battery health against a standardized end-of-discharge (EOD) time under predefined discharge conditions. This approach enables a consistent evaluation of battery performance and end-of-life (EOL). HEM is first established by statistically selecting the benchmark battery and its EOD time under the specified discharging parameter. We then develop a graph convolutional network-based HEM estimation model to apply HEM to target batteries. This model requires only partial discharge voltage segment data as input, facilitating efficient health monitoring and remaining life estimation. Using 90 batteries from a widely recognized public dataset, our results demonstrate that under the HEM framework, the battery life of long-range batteries is extended by up to 4.6\%, while short-range batteries are protected from the risk of overextending their lifespan compared to SOH. This validates HEM as a consistent metric for accurate health assessment, helping to prevent premature battery retirement or excessive use.