The structural integrity of composite materials in unmanned aerial vehicles (UAVs) is critical for ensuring flight performance, safety, and durability under various opera- tional conditions. These materials are subject to progressive damage accumulation due to external loads, fatigue cycles, and environmental factors, which can compromise their mechanical properties over time. This work proposes a damage classification system based on Fiber Bragg Grating (FBG) sensors and Artificial Neural Networks (ANN) to detect and classify structural damage in composite material parts of UAVs. The valida- tion of the methodology is performed in data from a composite wing section represen- tative of an unmanned aerial vehicle (UAV) structure. The system integrates thirty-two FBG sensors strategically placed to measure strain variations through wavelength shifts, capturing subtle deformations indicative of material degradation. The sensor data collected are processed to classify the material conditions into five damage states ranging from undamaged to increasingly deteriorated structural responses. The neural network model, designed as a multilayer perceptron (MLP), receives wave- length shifts as input, leveraging supervised learning to achieve accurate damage classification. The model is trained on a diverse dataset generated from controlled experiments to enhance robustness, where composite UAV materials are subjected to various stress conditions to ensure that the ANN can generalize damage classification across different failure modes commonly observed in UAV structures. 210 labeled experiments were used for training and validation, providing the model with a broad representation of structural responses. The results demonstrate that the proposed ANN-based approach achieves high classification accuracy, effectively distinguishing different levels of mate- rial degradation based on the FBG sensor data.
González et al. (Tue,) studied this question.