The exposure of carbon-fiber-reinforced polymers (CFRPs) to open-field conditions was investigated. Establishment of structure-property relations with nanoindentation enabled the observation of modification effects on carbon-fiber interfaces, and impact resistance. Mapping of nanomechanical properties was performed using expectation-maximization optimization of Gaussian fitting for each CFRPs microstructure (matrix, interface, carbon fiber), while Weibull analysis connected the weathering effect to the statistically representative behavior of the produced composites. Plasma modification demonstrated reduced defect density and improved nanomechanical properties after weathering. Artificial intelligence for anomaly detection provided insights on condition monitoring of CFRPs. Deep-learning neural networks with three hidden layers were used to model the resistance to plastic deformation based on nanoindentation parameters. This study provides new assessment insights in composite engineering and quality assurance, especially during exposure under service conditions.
Keywords: artificial intelligence; carbon fibers; composites; deep learning; impact behavior; interphase; nanoindentation; neural networks.