<p>In the "design-simulation-preparation-characterization" research framework for carbon nanomaterial-reinforced polyvinyl alcohol (PVA) composite fibers, molecular dynamics (MD) simulations play a core role. Using Materials Studio 2023 and LAMMPS, oriented composite models of carbon nanotubes (CNT), graphene (GN) with PVA, along with pure PVA control models, were constructed. Simulations included geometric optimization, NPT dynamic processes, and annealing treatments, followed by calculations of mechanical properties (Young's modulus, shear modulus, density) and Z-direction stress-strain curves. Confined shear simulations mimicked shear-stretching in wet spinning, with relative concentration (RC) analysis to evaluate defect formation, while cohesive energy density (CED) analysis assessed composite compatibility. These simulations guided experimental design (e.g., selecting functionalized GO and CNT via CED results), validated experimental observations through stress-strain curve comparisons, revealed reinforcement mechanisms (carbon nanomaterials regulate stress transfer, reduce defects, and promote PVA crystallization), and established a "simulation-guided, experiment-validated" research framework, providing theoretical tools and methodological references for interface engineering in polymer nanocomposites.</p>