Partial Solution-Guided Machine Learning Enhanced the Impact Resistance Analysis of Prestretched Graphenes

QH Ge and WP Zhu, ACS APPLIED NANO MATERIALS, 7, 20163-20175 (2024).

DOI: 10.1021/acsanm.4c02997

This study aims to investigate the effect of prestretching on a graphene's ballistic limit velocity under impact loading, which is the maximum impact velocity that graphenes can withstand without being penetrated by a bullet. Molecular dynamics simulations and machine learning were employed to analyze the ballistic limit velocity of graphenes under various prestretching conditions. A machine learning model was developed by integrating a partial solution into the loss function to expedite the determination of the ballistic limit velocity. The physical mechanisms of the prestretching effect under impact loading were explained based on the stiffness theory of corrugated plates. The study revealed that uniaxial stretching is superior to equibiaxial stretching in inducing the prestretching effect. The machine learning model effectively predicted the ballistic limit velocity of graphene at different tensions and bullet sizes. When the bullet radius is small, the ballistic limit velocity exhibits significant fluctuations with increasing tension, whereas it decreases monotonically with increasing tension when the bullet radius is large. Although the positive effects of prestretching diminish with an increase in the number of graphene layers, prestretching significantly reduces the interlayer spacing of the graphene Whipple structure. These results provide valuable insights into the application of graphene in engineering, particularly in enhancing its ballistic performance. The findings suggest that prestretching can be an effective engineering manipulation to improve the impact resistance of graphene-based materials, which can be crucial for designing more effective protective materials and devices.

Return to Publications page