Machine-learning-assisted multiscale modeling strategy for predicting mechanical properties of carbon fiber reinforced polymers

Carbon fiber reinforced polymers (CFRPs) possess light weight and high strength, making them highly attractive for various applications. However, the design parameter space of CFRPs is extensive, with the complex relationship between structures and mechanical properties. Traditional design methods that rely on trial and error or scientific intuition are laborious and expensive for achieving optimal properties of CFRPs. In light of this challenge, we proposed a machine-learning-assisted multiscale modeling strategy that can efficiently predict the mechanical properties of CFRPs. This strategy uses low-computational-cost machine learning (ML) models to replace traditional theoretical models and combines them with molecular dynamics simulation to predict the mechanical properties of CFRPs starting from resin molecules. Comparing predicted values with the proof-of-concept experiment and the existing experimental findings showed that the predicted values of the ML model are in good agreement with the experimental ones. This strategy can be a viable machine-learning-assisted solution to designing CFRPs.