Metal Additive Manufacturing (MAM), often known as metal 3D-printing with metals, is a ground-breaking technology that is especially promising for creating complex parts used in critical areas like aerospace. However, ensuring the reliability of these parts has been a challenge, slowing down the adoption of MAM for such important uses.
To tackle this, researchers within Mechanical Engineering have been using powerful X-rays to watch the metal as it’s being printed, which helps them understand and improve the process. But these X-ray experiments produce a massive amount of data, too much for humans to analyse by hand.
A team of experts, including Prof. Lee and Dr. Leung created AM-SegNet, a smart, lightweight neural network designed to quickly and accurately analyse the X-ray images from these experiments. The team trained AM-SegNet using a huge database of over 10,000 labelled images from top research facilities around the world.
The results have been impressive. AM-SegNet can analyze an image with about 96% accuracy in less than 4 milliseconds, and it’s more reliable than other advanced models. This means it can rapidly process the data, helping researchers get to the important insights faster.
By speeding up data analysis, AM-SegNet is helping to uncover the detailed physics of the AM process, which in turn helps manufacturers improve their methods and achieve more reliable results. It’s a significant step forward for real-time monitoring and quality control in metal additive manufacturing.
AM-SegNet is a game-changer, making it easier and faster to analyse data and gain insights, which could lead to more efficient and reliable manufacturing processes.
The source codes of AM-SegNet are publicly available at GitHub (https://github.com/UCL-MSMaH/AM-SegNet).
Professor Peter Lee
Principal Investigator
Dr Alex Chu Lun Leung
Principal Investigator
