Advanced X-ray radiographic imaging analysis for identifying water dynamics regime of polymer electrolyte fuel cells

Abstract

Operando X-ray radiography is widely used to evaluate liquid water distributions in polymer electrolyte fuel cells, yet conventional analyses typically condense pixel-resolved time series into spatiotemporal averages, potentially obscuring dynamically fluctuated water states. In this study, we present a computational spatiotemporal statistical framework to extract fluctuated information from sequential radiographic datasets of a cathode gas diffusion layer operating under steady conditions. Beyond mean saturation index, pixel-wise time-series descriptors, including skewness, excess kurtosis, Shannon entropy, frequency, autocorrelation and mutual information, are systematically evaluated to characterise distributional variability, periodicity, temporal persistence and spatial coordination of water fluctuations. The analysis reveals that regions with similar mean saturation can exhibit distinct fluctuation structures. By combining Shannon entropy and temporal autocorrelation, a two-dimensional statistical regime map is established to classify fluctuation patterns into four states. The proposed framework does not directly establish universal water-transport mechanisms, but instead introduces a fluctuation-oriented statistical perspective complementary to traditional mean-value analyses. This approach provides a reproducible and extensible methodology for comparative interpretation of operando radiographic datasets and may contribute to future data-driven diagnostic analyses.