A network developed by Russian engineers could dramatically cut the time needed to 3D print new structures.
Researchers from Peter the Great St. Petersburg Polytechnic University recently detailed a neural network that helped the team print metal structures. Engineers said current mathematical modeling processes can require a week’s worth of calculations even for simple structures.
Using the large-scale computer network, however, allowed the researchers to develop parameters for printing new structures, ensure effective melting of metal and check the stability of its welding.
In addition to quicker printing of prototypes, the team said the system can rely on previous data to speed up manufacturing of subsequent parts.
Officials next hope to move from manual data entry into the network to an automated, online system that receives data and churns out printing parameters while “learning continuously.”
“We believe that the new system will improve the quality of the parts and increase the speed of parameters development for further manufacturing," Oleg Panchenko, who heads the university’s Laboratory of Lightweight Materials and Structures, said in a statement.
