According to the University of Toronto, the researchers use machine learning to improve the 3D printing of metal. Under the direction of Yu Zou, a professor of materials and engineering at the Faculty of Applied Science and Engineering, the research team has developed a new framework that was mentioned as an exact inverse process optimization framework in the energy separation designated by Laser Crispected Energy Deposition (AIDD). The prediction of how the metal melts and consolidates to find optimal pressure conditions improves the new framework that is published in an article published in the additive manufacture, the accuracy and robustness of the finished product.
The researchers say that the approach can be used to produce higher-quality metal parts for industries such as aerospace, automotive, nuclear and health care.
“The more comprehensive introduction of directed energy separation – a large metal -3d printing technology – is currently hindered by the high costs for the search for optimal process parameters by experiment and error,” said Xiao Shang, first author of the new study. “Our framework quickly identifies the optimal process parameters for various applications based on the requirements of industry.”
“A major challenge of the 3D metal pressure is the speed and precision of the manufacturing process. Variations in the pressure conditions can lead to inconsistencies in the quality of the end product, which makes it difficult to meet the industry standards for reliability and security,” said Zou. “Another major challenge is to determine the optimal settings for printing different materials and parts. Every material obtanium for aerospace and medical applications or stainless steel for the core reactor-hats unique properties that require specific laser performance, scan speed and temperature conditions. The correct combination of these parameters via a large selection of the parameters of process parameters.
In order to cope with these challenges, AIDD works in a system with a closed circuit, in which a genetic algorithm-a method, which imitates natural selection in order to find optimal solutions, first suggests combinations of process parameters, the machine learning models then rate for pressure quality. The genetic algorithm checks them to ensure that they are optimal – and repeat the process until the best parameters are found.
“We have shown that our framework can identify optimal process parameters from customizable destinations in just one hour and predict geometries from process parameters,” said Shang. “It is also versatile and can be used with different materials.”
To develop the frame, the researchers carried out numerous experiments to collect their huge data records. The team is now working on developing an improved autonomous or self-driving metal 3D printing system that works with minimal human intervention, similar to autonomous vehicles itself. “By combining the most modern add-up methods with artificial intelligence, we want to create a new, controlled self-driving laser system with closed loops,” said Zou. “This system can capture potential defects in real time, predict problems before they occur and automatically adjust the processing parameters to ensure high -quality production. It will be versatile enough to work with different materials and sub -geometries, which makes it a player for the production of industries.”
In the meantime, the researchers of the University of Toronto will hope that an AIDED -AIDED will change process optimization in industries that are currently using metal 3D printing. “Industries such as aerospace, biomedical, automobiles, core and more would welcome such an inexpensive but precise solution to facilitate their transition from traditional production to 3D printing,” said Shang.
“By 2030, additive production will probably be redesigned in several high -precision industries,” said Zou. “The ability to correctly correct defects and optimize parameters accelerates the introduction.”