Artificial neural networks applying to detect damage in carbon fiber/epoxy composite

Artificial Neural Networks (ANNs) have emerged as one of the most useful tools in Artificial Intelligence (AI), being used in the most applications, such as engineering, economics, and health.Due to their great capacity for learning, adaptation, and generalization, ANNs can handle linear and nonlinear models that other methods are not capable to solve.Composite materials are increasingly used in critical and demanding applications, mainly due to its high specific strength and stiffness.However, the challenging of the current Structural Health Monitoring (SHM) methodologies includes identification, detection, and quantification of the damage and, also the prediction of the residual resistance/life of the structure.Therefore, this work proposes a methodology based on ANNs to detect damage in composite structures.Initially, the vibrationbased method was applied using Frequency Response Functions (FRFs) along with Principal Component Analysis (PCA).This tool seeks to reduce the dimensionality of the original data while maintaining its characteristics.Next, a multi-layer neural network was developed for detecting damage in composite plates made of Carbon Fiber Reinforced Polymer (CFRP).Finally, it is discussed the potentialities and limitations of the methodology for use in damage detection systems.