Résumé:
Detecting cracks in structures is very important to ensure their safety and reliability. In composite
materials like CFRP (carbon fiber-reinforced polymer), cracks can be difficult to detect, especially at
an early stage. Vibration-based methods are commonly used because they do not damage the structure
and can detect internal flaws.
This study investigates the use of artificial neural networks (ANNs) for the identification of crack
depth in carbon fiber-reinforced polymer (CFRP) composite beams based on vibration analysis.
Experimental modal testing was conducted on both intact and notched cantilever beam specimens to
obtain natural frequencies under various damage scenarios. Finite element models were developed in
MATLAB to simulate these configurations and validate the experimental data. The first three natural
frequencies were used as inputs to train a multilayer ANN model aimed at predicting crack depth.
Several network architectures were evaluated, with the model comprising 10 hidden layers
demonstrating more accuracy and high performance. The results confirm the effectiveness of the ANN-
based approach for crack depth prediction in composite structures.
