Durability Evaluation and Small-Sample Prediction of Ultraviolet Aging in Polyurethane Elastomers

UV aging significantly affects the long-term performance of polyurethane elastomers, while quantitative description and short-range extrapolation of degradation trends remain challenging under limited experimental data conditions. In this work, accelerated UV aging experiments were conducted on microcellular polyurethane elastomers with different densities under UVA-340 irradiation. Microstructural evolution during aging was systematically characterized by Fourier transform infrared spectroscopy, scanning electron microscopy, small-angle X-ray scattering, and thermogravimetric analysis, while mechanical properties including hardness, stress, strain, tensile strength, and yield-related parameters were monitored. To enable quantitative description of performance degradation trends, a gray prediction GM(1,1) model was employed to describe the evolution of hardness during UV aging. The model shows good fitting consistency for both materials and provides a basis for describing the measured hardness trend and illustrating short-range model-based extrapolation under limited-data conditions. In addition, Pearson correlation analysis and Kendall’s concordance test were applied to elucidate the statistical relationships and consistency among mechanical performance indicators. The results reveal coordinated evolution among fracture-related properties, whereas yield-related parameters exhibit weaker statistical coupling, suggesting partially decoupled mechanical-response trends during UV aging. The GM(1,1) analysis in this work is limited to Shore A hardness and is used to describe the hardness-evolution trend and short-range extrapolation under limited-data conditions. Overall, this work combines accelerated aging experiments, multiscale characterization, hardness-based gray-model analysis, and statistical comparison of mechanical indicators to provide a case-specific description of UV-aging behavior in microcellular polyurethane elastomers under limited-data conditions.