Phy-GCN: A physics-constrained graph convolutional network incorporating structural dynamic response features for structural damage identification and localization

Puntos clave

• Structural damage identification shows increased accuracy using a graph convolutional network approach.
• Key evidence indicates successful localization with notable dynamic response features in testing.
• The approach integrates physics constraints within a graph convolutional network for improved results.
• Enhances current methodologies for identifying damage in structures; external validation recommended.