ITA graduate develops AI-supported monitoring system to detect damaged carbon fiber
Deniz Sinan Yesilyurt wins a Young Talent Award from VDMA for his Bachelor’s thesis that used machine learning to recognize various fiber defects with up to 99% accuracy during thermal stabilization.
Winners of the VDMA Young Talent Award. Deniz Yesilyurt is depicted in the first row, far left. Photo Credit: VDMA
On Dec. 6, RWTH Aachen University Institut für Textiltechnik (ITA, Germany) junior engineer and Bachelor’s graduate, Deniz Sinan Yesilyurt, received the “Digitalisation in Mechanical Engineering” Young Talent Award initiated by the German Engineering Federation (VDMA) in Frankfurt am Main, Germany. He was awarded second place for his Bachelor’s thesis “Development of a Kl-supported process monitoring system using machine learning to detect fibre damage in the stabilisation process,” in which he developed an AI-based camera system to automatically detect and classify fiber defects in carbon fiber production in order to reduce production costs and increase inspection efficiency.
Carbon fiber production is considered to be a highly complex process. Thermal stabilization, for example, is a highly manual process in which up to 300 individual fiber strands (bundles of individual fibers) have to be monitored simultaneously to avoid fiber damage. Due to the size of the ovens and many fiber strands, fiber defects often cannot be identified at an early stage in industrial production, leading to fiber damage, and thus, downtime to remove the damaged fibers.
To find a solution to this, Yesilyurt considered the integration of artificial intelligence (AI) into the production process, as a means to mimic human vision by using camera systems to automatically complete predefined tasks. Yesilyurt attached a camera to the carbon fiber line that takes pictures of various fiber defects during production and collects them in a database. The AI within the camera’s information technology system is then able to evaluates any fiber defects by assigning the images to predefined reference defects. According to the engineer, the AI system is able to recognize various fiber defects with a classification accuracy of 99%. The developed process can also be used in other areas that produce chemical fibers.
This project was part of the larger Carboscreen research program reported by CW, “CarboScreen project to develop multimodal, sensor-based monitoring of carbon fiber production process.” For more information, download the thesis paper here.
The VDMA awarded the prize to a total of four theses from different universities from Germany, Austria and Switzerland.
View this link for the original article.
Related Content
-
Materials & Processes: Fibers for composites
The structural properties of composite materials are derived primarily from the fiber reinforcement. Fiber types, their manufacture, their uses and the end-market applications in which they find most use are described.
-
Infinite Composites: Type V tanks for space, hydrogen, automotive and more
After a decade of proving its linerless, weight-saving composite tanks with NASA and more than 30 aerospace companies, this CryoSphere pioneer is scaling for growth in commercial space and sustainable transportation on Earth.
-
One-piece, one-shot, 17-meter wing spar for high-rate aircraft manufacture
GKN Aerospace has spent the last five years developing materials strategies and resin transfer molding (RTM) for an aircraft trailing edge wing spar for the Airbus Wing of Tomorrow program.
