Terahertz-based optical parameters analysis and quantitative inclusion defects detection in glass fiber-reinforced polymer laminate

Abstract

Inclusion defects are often introduced in the manufacturing process of glass fiber-reinforced polymer (GFRP) components, which can be effectively identified by Terahertz (THz) technology. However, accurate measurement of the size and location of defects for engineering applications remains a challenge. In this study, based on the optical parameters of GFRP laminate, a quantitative detection of inclusion defects was conducted. For defect area measurement, a defect area measurement algorithm based on super-resolution generative adversarial network (DAMSRGAN) was proposed, enhancing measurement accuracy by employing generative adversarial networks to improve image resolution. The final quantification of defect area was achieved through a combination of threshold segmentation and blob analysis. Compared to traditional methods for characterizing defect areas based on raw low-resolution time-of-flight tomography (TOFT) images, the proposed algorithm effectively enhances measurement accuracy. For defect depth measurement, the influence of the number of layers and ply angles of GFRP laminates on THz optical parameters was studied, revealing an approximate linear relationship between the number of layers and refractive index of GFRP laminates. Based on this relationship, the refractive index of the tested GFRP sample can be estimated, thereby eliminating the need to remove it from the assembled structure for optical parameter measurement. Furthermore, defect depth information can be calculated based on the estimated refractive index, enhancing the convenience of detecting GFRP defect depth using THz technology. This study provides a valuable supplement for the accurate and convenient measurement of inclusion defects in GFRP components using THz technology.