Enhancing automatic inspection and characterization of carbon fiber composites through hyperspectral diffuse reflection analysis and k-means clustering
The International Journal of Advanced Manufacturing Technology(2024)
摘要
Numerous industries utilize carbon fiber composites (CFC) for their exceptional strength-to-weight ratio and stiffness. However, inherent manufacturing defects such as voids and delamination can undermine the material’s structural integrity and performance. This study introduces an advanced imaging technique employing hyperspectral imaging (HSI) to effectively detect and characterize flaws within CFC materials. HSI provides high-resolution spectral data, enabling precise analysis of material properties. Initial observations indicate a distinctive peak variability in dispersed reflectance spanning 440 nm to 600 nm for both pristine surfaces and defective regions of CFC sheets. The proposed methodology entails the automated characterization of CFC through a combination of HSI and an advanced clustering technique, k-means clustering (k-mc). The application of k-mc facilitates rapid and accurate categorization of defect locations. Statistical analysis reveals mean and standard deviation (SD) values of 0.34 and 0.33, respectively, for normal CFC surfaces, compared to 0.43 and 0.29 for cracked CFC sections. These discernible variations enable precise differentiation between defect-free and defective CFC specimens. By employing spectral signatures at 445 nm, 546 nm, and 585 nm as optical markers, the study accurately measures fracture penetration depths. Implementation of this approach generates 3D-resolved images, offering a comprehensive visualization of CFC imperfections. The proposed methodology presents a more automated and objective strategy for defect identification and categorization. This unique approach holds significant potential for industrial applications, particularly in scenarios necessitating efficient and precise evaluation of numerous CFC components.
更多查看译文
关键词
Carbon fiber composites,Automatic inspection,Non-destructive,Hyperspectral imaging,K-means clustering
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要