AI technology for virtually manufacturing metal surfaces has been developed, contributing to increased productivity by reducing quality variations caused by process operators in manufacturing sites.

Professor Im-Doo Jeong's team at the Department of Mechanical Engineering at UNIST announced on the 15th that they have succeeded in developing a technology to virtually manufacture '3D metal surface shapes expected according to metal processing conditions' based on actual data.

This technology is an advancement of AI technology into the 'Direct Energy Additive Manufacturing (DED) process,' a 3D printing technique that is useful for manufacturing large metal parts such as rocket components or repairing broken parts that are difficult to repair because they are no longer processed.

However, there has been a problem in that controlling surface characteristics that affect fatigue properties is difficult, resulting in high process development costs when using expensive materials such as titanium for non-experts.

With a technology developed by Jeong Im-doo's research team that utilizes AI to virtually generate surfaces expected based on DED process conditions, it is expected that even unskilled workers will be able to easily obtain desired surface characteristics.

AI learns surface scan images based on laser output, powder spray speed, and scan speed, and creates expected virtual 3D surfaces for arbitrary process inputs. We quickly created metal surface images expected under various process conditions in seconds.

The research team stated that the characteristics of the metal surface manufactured using the actual identical process were well represented, and that the surface manufactured under the process conditions recommended by AI also exhibited excellent microstructural properties.

Meanwhile, the research results were published in 'Virtual and Physical Prototyping,' an international academic journal ranked within the top 5% of the world's JCR rankings in the manufacturing sector.

Professor Im-Doo Jeong of the Department of Mechanical Engineering at UNIST, who led the research as the corresponding author, stated, “In the manufacturing industry, quality is often heavily dependent on the skill level of workers, and the absence of workers with know-how can disrupt operations.” He added, “The more technologies like the virtual manufacturing AI developed in this study are created, the more they can reduce quality variations caused by process workers and contribute to the ultimate improvement of productivity through digitalization.”