Speaker
Description
While basic beam parameters can be determined quickly using an optics model for the Proton Storage Ring (PSR) at LANSCE, some parameters, like effective beam current, beam sizes, and the impacts of RF bunchers, normally require multi-turn beam dynamics simulations that are often too long for the control room. In this effort, we present an ML surrogates based on PyORBIT simulations of the PSR to allow real-time feedback for parameters requiring 1715-turn simulations as well as an improvement to the optics parameters We further demonstrate an AI-assisted workflow that streamlines Monte Carlo simulation generation, post-processing, data analysis, model building, and training, enabling rapid iteration of surrogate models. This rapid prototyping capability, combined with fast surrogate inference, enables a model-based diagnosis for real-time operational decision-making in the CCR.
| I have read and accept the Privacy Policy Statement | Yes |
|---|