Year: 2024
Type: conference-paper
Abstract: Reliable monitoring for fast manufacturing processes, such as resistance spot welding is quite challengingin the face of uncertainties, process drift and long model evaluation times. It is however an essential basisto perform adequate control and receiving good output quality consistently. In the face of noisy data andslow-to-evaluate finite element models, that are magnitudes slower than the process itself, grey-box modeloffer a solution that is well suited for uncertainty quantification and addressing process drift. In particular,the authors look at a nested multifidelity model, which is easy to implement into a control loop and offersbenefits of incorporating prior knowledge gained from a numerical model of the process as well as liveprocess data. A simple electric heating equation is studied to compare the predictability and performance ofsingle-fidelity models with two high-fidelity models. The model is further affected by process drift as wellas measurement and input uncertainties.
Source: Lirias
Authors: Dodt, Miriam B., Bogaerts, Lars, Persoons, Augustin, Faes, Matthias G.R., Moens, David
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: European Commission
OpenAlex ID: W6911726693
FWCI:
Cites:
Cited by:
Related to: