Small-Sample Battery Capacity Prediction Using a Multi-Feature Transfer Learning Framework

Batteries, Journal Year: 2025, Volume and Issue: 11(2), P. 62 - 62

Published: Feb. 7, 2025

The accurate prediction of lithium-ion battery capacity is crucial for the safe and efficient operation systems. Although data-driven approaches have demonstrated effectiveness in lifetime prediction, acquisition lifecycle data long-life lithium batteries remains a significant challenge, limiting accuracy. Additionally, varying degradation trends under different operating conditions further hinder generalizability existing methods. To address these challenges, we propose Multi-feature Transfer Learning Framework (MF-TLF) predicting small-sample scenarios across diverse (different temperatures C-rates). First, introduce multi-feature analysis method to extract comprehensive features that characterize aging. Second, develop transfer learning-based framework, which leverages pre-trained models trained on large datasets achieve strong performance data-scarce scenarios. Finally, proposed validated using both experimental open-access datasets. When small sample dataset, predicted RMSE error consistently stays within 0.05 Ah. results highlight MF-TLF achieving high accuracy, even with limited data.

Language: Английский

Bio-inspired hybrid materials for sustainable energy: Advancing bioresource technology and efficiency

Materials Today Communications, Journal Year: 2025, Volume and Issue: 46, P. 112647 - 112647

Published: April 25, 2025

Language: Английский

Rational Design of Yolk–Shell Fe₇S₈@C-N for High Rate and Long Cycle Li-Ion Batteries

Nano Letters, Journal Year: 2025, Volume and Issue: unknown

Published: May 5, 2025

Fe7S8 with large capacity shows high potential for Li-ion batteries, while it still suffers volume expansion, resulting in fast fading. Herein, a novel yolk-shell structural Fe7S8@C-N is rationally designed, which the N-doped carbon layer superior mechanical flexibility enables one to accommodate expansion of core and promote its electronic transportation. Besides, surface porous morphology believed facilitate electrolyte infiltration diffusion as well. Therefore, this modified electrode exhibits lower expansivity (∼28.0% vs ∼87.4%), smaller voltage hysteresis, higher conductivity (1.6 × 10-2 S/m) better Li-diffusivity (1.09 10-12 cm2/s) than pure powder; thus cyclability (458 mAh/g 121 after 150 cycles) rate-capability improvement (546 125 at 2000 mA/g) can be achieved. Such design strategy easily extended other conversion or alloying type materials advanced energy storage.

Language: Английский