Automatic detection of impact damage in carbon fiber composites using active thermography

• Published in: Infrared physics & technology 2013-05, Vol.58, p.36-46
• Main Authors: Usamentiaga, R., Venegas, P., Guerediaga, J., Vega, L., López, I.
• Format: Article
• Language: English
• Subjects: Accidental impacts; Carbon fibers; Defects; Degradation; Heating; Impact damage; Infrared image processing; Infrared inspection; Inspection; Optical step heating thermography; Phasegram; Thermography
• ISSN: 1350-4495; 1879-0275
• DOI: 10.1016/j.infrared.2013.01.004

► A novel method is proposed to automatically detect impact damage in CFRP using active thermography. ► Non-uniform background caused by non-uniform heating is removed. ► Peaks and edges in the images are clustered, and regions corresponding to impact damage are detected. ► A test set with different number of plies, types of cores, and damage caused by energies from 6J to 50J is used. ► The number of true positive detections is 100%, and the number of false negatives and false positives is null. Accidental impacts can severely reduce the structural strength and stability of composite materials, which can lead to severe consequences due to the degradation of the mechanical properties of components designed to perform for decades. Because accidental impacts are difficult to avoid, robust and reliable inspection methods to detect impact damage are required. Many methods have been proposed recently. However, most of them require an experienced technician to analyze the data, which leads to a significant decrease in manufacturing productivity. This work proposes a method to automatically detect impact damage in carbon fiber composites using active thermography. The proposed system detects defects caused by impact damage in the infrared images without human intervention. Impact damage detection is performed using a robust method based on an active thermographic inspection. Thermographic data is preprocessed to improve signal-to-noise ratio and to remove non-uniform background caused by non-uniform heating. Then, peaks and edges are identified and clustered, and regions corresponding to impact damage are detected. The proposed procedure has been applied to three specimens that contain 6 and 12 plies, different types of cores, and damage caused by energies from 6J to 50J. All defects are detected correctly.