Capacity Prediction of Lithium-ion Batteries with Regeneration Phenomena Based on Hierarchical Modal Analysis

Capacity prediction plays a critical role in battery health management. In recent years, the regeneration phenomenon has been considered to increase the accuracy of capacity prediction. Against this background, this method innovates upon the existing hybrid technique by integrating complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and adaptive neuro-fuzzy inference system (ANFIS). The advancement lies in the selective application of variational mode decomposition (VMD) to the intrinsic mode function identified by CEEMDAN, which exhibits regeneration features. This targeted re-decomposition allows the significant fluctuations in these components to be captured more effectively. The effectiveness of the proposed approach has been validated using NASA's lithium-ion battery aging datasets, showcasing an improved prediction accuracy and a better explanation for the capacity regeneration phenomenon. The research presents a significant contribution to the capacity prediction of lithium-ion batteries, providing a robust predictive framework that could be vital for the advancement of battery management systems.