A graph neural network-based computational framework for predicting mechanical behavior of continuous carbon fiber/PEEK composites: from microstructural features to macroscopic applications

Key Points

• Mechanical behavior predictions are enhanced using graph neural networks, demonstrating significant potential.
• The approach connects microstructural features to macroscopic applications in composite materials.
• Graph neural network framework integrates deep learning with material science for comprehensive analysis.
• This framework supports improved predictions for carbon fiber composites, potentially transforming material design.