With the aim of moderating the consumption of traditional fuels and carbon

Capacity attenuation presents a significant challenge in the application of lithium-ion batteries, complex due to the disintegration mechanisms at play. This study aims to address the intricacies of capacity prediction using a machine learning approach, specifically implementing the Whale Optimization Algorithm-Extreme Learning Machine (WOA-ELM) model. The model exhibited an R2 value of 0.9999871, indicating highly accurate predictions of the specific capacity changes within battery cycles. Furthermore, this research identifies the key factors influencing battery capacity, effectively overcoming the black box limitations inherent in traditional machine learning models by employing interpretable techniques. The study elucidates the relationship between these factors, structural damage in electrode materials, and overall battery failure during cyclical processes, thus providing valuable insights into enhancing battery performance and informing future material modifications. The findings contribute significantly to advancing contemporary battery research and practical applications, ultimately supporting the transition to more sustainable energy solutions.