About this Abstract |
Meeting |
TMS Specialty Congress 2024
|
Symposium
|
2nd World Congress on Artificial Intelligence in Materials & Manufacturing (AIM 2024)
|
Presentation Title |
Real-time Predictions of Distortion and Residual Stress Resulting From Weld Sequences Using Machine-learning Algorithms |
Author(s) |
James Sobotka, Matthew Robinson, Jake Janssen |
On-Site Speaker (Planned) |
James Sobotka |
Abstract Scope |
Physics-based simulation is a common tool in the fabricated components and structures community. Unfortunately, these simulations do not account for real world manufacturing variation, i.e., geometric ill-fit, that may dictate or drive a process change. Using advances in artificial intelligence and new machine learning frameworks, we now have an opportunity to train a real-time algorithm that can assess the input variation to a process and to update in real-time a process based on the geometry presented, even though it deviates from CAD. In this talk, we outline a framework to enable an intelligent robotic system to visualize a presented condition and leverage learning from physics-based simulations and ongoing production history to generate optimal process plans. These developments take advantage of both new approaches for machine learning based frameworks as well as the ability to execute at the rate of production, leading to improved operational efficiencies and better performing fabricated products. |
Proceedings Inclusion? |
Definite: Other |