Local elasticity assessment of unidirectional fiber-reinforced polymer composites through impulse excitation and machine learning

This study presents a novel methodology that integrates the Impulse Excitation Technique (IET) and machine learning (ML) to predict local elastic properties within isolated regions of unidirectional polymeric composite plates. The proposed model incorporates fiber volume and plate thickness as input parameters and leverages the first resonance frequencies of the local region at different fiber orientations, thus accounting for the composite's anisotropy. Regression results from the deep neural network (DNN) model demonstrate robust prediction performance across all output targets in both testing and training datasets, with R2 coefficients surpassing 0.9. The model exhibits particularly strong performance in predicting Young's moduli. Additionally, each output objective shows sensitivity to a unique balance of input parameter weight factors for achieving optimal ML predictions. Moreover, a parabolic trend in the weight factors of fundamental frequencies at different orientations is observed as the rigidity of composites changes. Lastly, a comparative study between carbon-fiber and glass-fiber composites highlights the variations in fiber volume and elastic constants, emphasizing the effectiveness of the proposed model in accurately predicting material properties.