Automated Fiber Placement (AFP) machines can be susceptible to tow-tape defects such as gaps and overlaps. These can reduce strength between 7% and 32%. Automated inspection is not fully effective, and current AFP systems rely heavily on visual-based inspections of each ply layer to detect and correct these tape defects.
In a method developed at Langley, the latent heat of the item under fabrication is used to create a thermal image of a just-completed tape bond. The image is then analyzed to detect anomalies in real time. The defect data can be used in a feedback process to guide the bonding operation and tag the defect location for subsequent inspection.
Image processing is a key element of successfully implementing the process. The image process technique used not only reduces processing resources (such as CPU usage, memory, etc.), but also allows for a number of standard time-based analysis algorithms, typical of flash thermography, to be applied to the data (the reconstructed sequence).