Sunday, December 4, 2022
HomeCommentaryUsing machine learning to accelerate composites processing simulation

Using machine learning to accelerate composites processing simulation

Manufacturing fiber-reinforced polymer is no mean feat. There are workmanship issues, naturally. However, science is also a part of the process. There are highly delicate processes, involving heat and mass and thermo-nuclear transitions. Though experience does count in knowing how to proceed through each part of the process, including how much time should be allowed, a new dynamic has recently been introduced. Because humans, even experienced humans, can be fallible, specialized digital simulation tools have been sought out as a means to oversee some of the process, reducing the chance of human-caused error. Current implementation practices are not without issues. Currently, the part being made and the tool it is being made for can be thoroughly analyzed using an 3D FE simulation tool. However, the process can be one of weeks. Now, in a bid to do better, machine learning models are being considered as a way to replace FE simulation tools. These machines are fast-learning. A model that has learned from an FE simulation tool is capable of replicating the effort of its teaching equipment, yet relaying the same process at a much quicker speed.

Key Takeaways:

  • Finite element (FE) simulation tools such as ABAQUS and ANSYS have advantages over old-fashioned trial and error.
  • 3D FE analysis is costly and time-consuming, so reduced-order FE is often used instead.
  • A new alternative is to use FE models to automatically generate a lot of data that can then train machine learning models.

“Current industrial implementation of this approach relies on conducting costly 3D FE analysis of the part and tool”

Read more: https://www.compositesworld.com/articles/using-machine-learning-to-accelerate-composites-processing-simulation

RELATED ARTICLES

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular

Recent Comments