A novel fractional calculus modeling and physics-informed machine learning study on dynamic performance of hybrid flax/basalt fiber-reinforced composite

Hybrid flax/basalt fiber-reinforced polymer (F/BFRP) offers promising mechanical improvements, but their dynamic properties under low-frequency vibration remain unclear. This study investigates how the basalt/flax fiber hybrid ratio affects the anisotropic dynamic behavior of the composites. The outcomes of the tests reveal that the storage modulus and loss factor of F/BFRP exhibit a nonlinear increase as both frequency and basalt fiber content increase. The loss factor demonstrates a similar nonlinear rise with frequency and flax fiber content. In pursuit of an accurate depiction of F/BFRP’s dynamic viscoelastic behavior, we present two complementary modeling approaches: a fractional calculus model offering a theoretical foundation and a novel Physics-Informed Neural Network (PINN) model that leverages neural network flexibility while incorporating the same physical principles. While the fractional calculus model excels in interpretability and analytical solutions, the PINN model prioritizes high accuracy and generalizability, particularly for data-rich scenarios. Both models demonstrate strong performance in their domains. Users can choose the approach that best suits their specific needs, whether they prioritize theoretical rigor or enhanced predictive power.