Abstract

This study presents an enhanced spectral framework for early fault detection in lithium-ion batteries, focused on analysing voltage and current fluctuations in silicon–carbon half-cells. The method integrates fast Fourier transform, continuous wavelet transform, cross-wavelet transform, and wavelet transform coherence to extract time–frequency features associated with degradation. Results reveal the emergence of low-frequency spectral modulations and phase coherence losses that precede critical failure events. The approach was validated across three independent silicon–carbon cells, demonstrating high reproducibility of fault indicators. Given its low computational footprint and diagnostic accuracy, the proposed method shows strong potential for implementation in battery management systems for predictive maintenance and real-time health monitoring.