How AI and Machine Learning Help Manufacture Components
Scientists at the Institute of Vehicle Systems Engineering at the KIT have developed a program that combines process simulation with machine learning and AI. This allows the algorithm to recognize patterns and estimate the manufacturability of a component.
What is currently being developed in Karlsruhe is called the "virtual AI process expert” by the scientists. Using machine learning and AI, they teach the algorithm to recognize patterns and thus estimate whether a component can be produced with the selected geometry or not.
"It's like putting the know-how and intuition of many engineers into one program and using it at the touch of a button," explains Clemens Zimmerling from the Institute of Vehicle Systems Technology at the Karlsruhe Institute of Technology (KIT), which developed the program. This virtual process expert is to be used in the design of lightweight structures made of fiber composites - both to quickly optimize the component and its manufacturing process.
Early Feedback on Manufacturability
The research project, funded by the Baden-Württemberg Ministry of Science, Research and the Arts, is primarily concerned with understanding and accelerating interpretation processes. The problem so far: Although simulations of the manufacturing process can be used to perform a virtual assessment of the manufacturability of components, these are very complex and time-consuming. For example, the successive optimization loops multiply the computing time. "The feedback we have received from developers is that they prefer a sound estimate at an early stage to a very precise one at a later stage," Zimmerling says. This is where the "virtual AI process expert" comes in. "If you use it in the early stages of component design, you get early feedback on manufacturability and an assessment of how to alter the manufacturing process. Thanks to the AI methods, it is possible to exclude unsuitable process variants from the optimization calculation and concentrate on the most promising variant for the relatively complex simulations," explains Zimmerling.
"Digitalization in particular is a major driving force behind lightweight design. It is especially important to jump on the bandwagon now and discover the possibilities of AI or machine learning for yourself and use them profitably - otherwise someone else will do so," says Dr. Wolfgang Seeliger, Managing Director of Leichtbau BW. Especially the shorter time-to-market is an enormous lever to create added value. “The complex simulations and optimization loops in lightweight design can only be reproduced digitally. If these are taken into account right from the start of product development, the full lightweight potential can be exploited," adds Seeliger.
Lower Costs for Small Batch Sizes
"By using our program, tooling and manufacturing costs can be reduced. In development, trial error tests are minimized, so that error-free production of the components can be achieved more quickly," says Zimmerling. In addition, the program would make it possible to realize small batch sizes more cost-effectively, in which, for example, changing components are produced on the same production line. This is especially interesting for industries with high derivatization, such as the automotive sector, mechanical engineering or aerospace.
This article was first published by konstruktionspraxis.
Original by Juliana Pfeiffer / Translation by Alexander Stark
This article is protected by copyright. You want to use it for your own purpose? Infos can be found under www.mycontentfactory.de (ID: 45901440)