基于卷积神经网络的石英纤维复合材料损伤缺陷太赫兹智能识别

利用太赫兹时域光谱对石英纤维复合材料(quartz fiber reinforced polymer, QFRP)内部分层缺陷进行检测,通过搭建一维卷积神经网络模型,实现不同位置和不同深度损伤缺陷的准确识别,验证结果准确率在90%以上.根据识别结果构建复合材料的缺陷检测图像与实际太赫兹成像图结果一致,且具有高清晰度和对比度.太赫兹技术结合卷积神经网络能够实现非极性材料的智能识别.

Intelligent identification of damage defects in quartz fiber composites using terahertz technique based on convolutional neural network

In this paper, terahertz time-domain spectroscopy was used to detect the delamination defects in quartz fiber reinforced polymer(QFRP), and a one-dimensional convolutional neural network was built to realize the accurate identification of damage defects at different positions and depths, and the accuracy of the verification results was more than 90%. The defect detection image of the composite constructed according to the recognition results has high definition and contrast, which was consistent with the actual terahertz image. Terahertz technology combined with convolutional neural network can realize the intelligent recognition of non-polar materials.